Conflat P - in 4mm Scale by Filament 3D Printing ?

Cleaning Up the CAD Model Further

Continuing with SketchUp Make 2017, having been rather critical of its importing and always repairing, it is actually performing a reasonable job of importing the Conflat P STL file for modifying the design of the wagon.

The repairs are being forced due to the loss of accuracy when converting to STL and again when re-creating the CAD model as I mentioned in my previous post on printing.

However it is best to avoid importing the STL file directly because SketchUp is very slow, the older 3ds format imports much faster, free converters are available on line.

If using a fast PC then the STL file can be imported directly.

Scaling the STL file from representing the full-size wagon to 40mm:ft (working at 10x the 4mm scale is safer in SketchUp) is best done with another application such as Netfabb prior to importing into SketchUp, otherwise SketchUp is likely to be scaling a damaged import making editing more unstable.

The detail of the W-irons and suspension has been simplified to improve stability of the CAD model whilst reducing the amount of detail which is unlikely to print well.

Changes applied so far.

1. Removed all the bolt head detail.
2. Simplified the leaf spring mountings, they got badly damaged when imported.
3. Rounding up or down w-iron and suspension dimensions to two decimal places, rather than to 6 decimal places.
4. Reinforcing the W-Iron attachment to the solebar
5. Reinforcing the mounting of the inboard brake shoe arms.

To-do

1. Modify the axle boxes to accept Peco top hat brass bearings.
2. Insert mountings for couplings.

CAD Models

W-iron and leaf spring assembly separated from the chassis for modifying. Left end of spring as simplified, right end as original.




W-iron and leaf spring assembly, initial modifications completed, both configured as SketchUp components until attached to the wagon frame.




Overall views of the wagon including the brake lever which is to be printed as a separate part.






Generating the STL File

The CAD model in SketchUp passes the testing by Inspector.

But loading into IdeaMaker - there are non-manifold edges, and a hole.

Here is a rather good explanation of manifold and non-manifold edges that I just found to illustrate these, I wish I had found this a year ago.

https://mimaki.com/support/faq/operation/entry-394743.html?iframe=true&width=100%&height=100%

Importing into Netfabb confirms this, and highlights the position of the hole.

Back to SketchUp, yes it has a surface on a W-Iron that shows as a shimmering face.

Inspector sees nothing wrong with it.

So delete the erroneous face.

After a few attempts I managed to get SketchUp to replace the 'shimmering' face with a solid face that wasn't reversed, that would also have caused an error in the STL file, orientation matters.

Useful video, "Netfabb Basics Part 1: Analysis and Mesh Repair"




Reloaded into Netfabb, still have another edge issue, too small to be visible.

Into IdeaMaker which still dislikes this error, auto-repair fixes it, but on reloading it is back again!

How flakey is the STL format?

Another STL File Format Limitation

Netfabb gives a clue, the nature of the STL file format means that an edge could break again when exporting even if repaired within Netfabb. Lo and behold reloading the repaired file back into Netfabb the repair had to be applied again.

I have experienced this before with IdeaMaker but had not worked out why this was happening.



However if the export is 'optimised', then all is okay.



The 'repaired and optimised' STL file now loads into IdeaMaker without issues.

The Workflow

So, time to summarise the workflow from downloaded STL to delivery to the printer as G-Code, as by now I am starting to lose track, so I'd best write it down now.

Import/export is from/to local disc.

Workflow Summary

1. Acquire the STL file of the item to be converted.
2. Netfabb - Import, initial scale (e.g. to 40mm scale), repair if necessary, export again.
3. SketchUp Make 2017 - Import and cleanup (remove all the surplus triangles), export.
4. IdeaMaker - Import STL file. If errors, stop and use Netfabb (5), otherwise to (7)
5. Netfabb - Import STL file, identify repairs that can be done in SketchUp back at (3), otherwise repair with Netfabb and save optimised if necessary.
6. IdeaMaker - Import the STL file which should be okay, if so scale down to 4mm scale and slice.
7. IdeaMaker - send G-Code to the printer. If prints okay, all done, otherwise (8)
8. SketchUp - update model, export STL and go back to step (4)

Simple eh!

Only kidding, if I did not document this I will have forgotten by next year.

All I have to do now is start printing sample parts before I commit to a 10 hour print run.

Jim

 

Hi Andy, not sure what you mean by body as the wagon is basically a plate wagon underframe with end panels and mounting plates for containers.

Jim

 

Well Andy, the original file scaled down to 4mm scale wiil have far too much detail to be printable with a fillament based printer, even after the slicer ignores details smaller than a certain unspecified size.

This is partly why I wanted to re-create an editable CAD model, which as it turns out is best done with SketchUp afterall.

Editing an STL file even by cropping would not prove much as I need to work from a stable CAD model to convert it to a layout resilient 4mm scale model.

What I need to know next is how well the present design prints and adjust it to suit the printing process as regards to shrinkage and how accurately the layers match the Z height dimensions, it can be quite noticeable if the print is +/- 0.2mm on e.g. a 0.5mm thick or high part.

Jim

 

Much time has been spent to find the magic combination of parameters to use the following materials, PLA for the printed part, PVA for a water dissolvable support medium, the nature of the item to be printed is too delicate for just pulling away supports printed in PLA.

So rather than bore you and me even more than I am with this to date, hopefully here is a less boring summary of the last few weeks of aggravation, the world of filament printing with a 0.2mm nozzle is very different to using a default 0.4mm nozzle.

Material Change

Stringing seems to be preventing progress with development, so a change of material should be considered.

The current material HatchBox PLA+ Pro has a shiny surface, so let’s look at a matte finish PLA as an alternative, the potential advantages are

1. They seem to string less than smooth or shiny surface filaments
2. This could be used unpainted to blend in with 1950s to 60s rolling stock.
3. Likely to be easier to paint than a smooth surface.

A matte black that appears to be favoured in reviews is the

ERYONE Matte PLA Filament 1.75mm, 3D Printer PLA Matte Filament, +/-0.03mm, 1kg(2.2lbs)/Spool, Matte Black S

Available on Amazon

https://www.amazon.co.uk/dp/B08JGP1JFB?ref=ppx_yo2ov_dt_b_fed_asin_title&th=1

So having bought a black and a white (latter for development and other tasks) it was time to run through determining the necessary slicer parameters to obtain a good print.

As usual I gave both of these PLA spools 4 hours drying at 50C, rotating every hour, 180,90,180 degrees before use, bagged the black, white into the printer to start testing.

Initial Results

First print with estimated parameters from earlier work with the HatchBox PLA using my retraction test tower was very good, string free, however then I started running into flow problems and failed prints, been here before.

It quickly became obvious that this was not nozzle blockage but an extruder feed issue, the drag from the filament was too high for the extruder drive which would then grind its way into the filament from which there is no return, like spinning car tyres in mud or snow, can't get out of it.

But why?

When unloading the filament by dragging the filament out by hand the ground in notch was clearly visible and the filament was not jammed in the hot end, however the drag from the tube which feeds the filament to the extruder was very high. Couple this with the higher force required to squirt the filament through a fine nozzle was just too much for the extruder to keep it moving. The extruder motor is strong so it just grinds its way into the filament, clogging up its serrated drive wheel in the process.

Matte and filled filament material (e.g. stone texture effect) can be quite rough, so swapped out the feed tube in the printer with a larger bore tube, 2.5mm instead of the standard 2mm, no difference.

Moved the spool to my external spool holder to straighten the path between spool and printer, no joy, eventually the reason for the jamming revealed itself.

As the filament was being pulled from the spool it was burying itself between windings in the next layer down, the significantly extra friction was causing the extruder to slip and grind away at the filament. By easing the filament path further by allowing it to curve over the top of the spool instead of from underneath and through a feed tube to the printer the feed issue was fixed.

This means that the filament is in the open, I'd rather have it enclosed to keep it dry and dust free so another option is required.

Modified Spool?

A standard '1kg' spool of 1.75mm diameter filament holds approximately 330m of filament, but, a spool could be modified to only hold one or two layers at near the outer diameter of the spool sides. This would eliminate the burying into the next layer, increase the ability of the filament to rotate the spool by pulling at a larger radius of action and the spool would be much lighter. Just rewinding onto a another spool is not easy as it needs to be done warmed so that it curves to match the core curvature of the spool, however if wound onto a larger diameter core it would readily grip it. This is a last resort option and would certainly hold enough material for a Conflat chassis print or two.

 

Not a daft question Andy, this first issue was due to the high friction surface of matte filament causing excessive drag in the filament feed path and getting ground into by the serrated drive wheel of the extruder, however, more jams for another reason coming up in the next exciting episode of "How crazy is Jim Freight?"

Jim

 

Humidity Controlled Dispenser

Resorted to placing the spool into a Kingroon dispenser box and this resolved the feed path friction problem and keeps the filament dust free and as a bonus in a humidity controlled container.

This has been modified by adding a support holster for the included hygrometer, the original transparent adhesive pad does not stick well to the container and besides if it did how would you change the button cell when it goes flat. Of the 4 of these I have bought for humidity sensitive materials only one cell died within a month of purchase, others have lasted at least 6 months so far.



Available from Amazon, not very cheap and the drilling for the support shaft does not take into account the asymetrical shape of the top of the container so it is skewed, however it works fine for me. You could probably make one for half the cost but my time is the more finite resource, at least my pensions trickle top up my cash, but not my aging

The included sillca gel sachets seem to last well, the original boxes bought nearly 6 months ago are still keeping the contents below 20% RH, and two are supplied with each box, I expect they can be dried again for re-use.

When using the top exit don't forget to plug the connector at the base of the box!

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

The included feed tube is far too stiff and short so that has been replaced with an 'ultra-smooth' 2.5mm bore PTFE tube.

https://www.amazon.co.uk/dp/B0D1CLBP4W?ref=ppx_yo2ov_dt_b_fed_asin_title&th=1

The same seller also sells the same bore of tube in a 5m length with a cutter which seems to be a lot cheaper than this but, having bought both, this 'ultra-smooth' does feel smoother IMO.

https://www.amazon.co.uk/dp/B0D1CHKF5R?ref=ppx_yo2ov_dt_b_fed_asin_title

Machine Configuration

As a reminder I am using a Raise3D E2, which has two independent direct drive extruders on a common gantry, an enclosed printing space and a heated bed.

The left hand extruder to print the required part in PLA and the right hand extruder to print the support material, e.g. PVA.

Materials

1. Eryone Matte PLA black and white.
2. Multicomp (Farnell) PVA - water dissolvable support material

Known issues with the materials

PVA Support

1. Does not like to stick to anything
2. Likes to string at every possible opportunity, if it's warm enough to print it will happily ooze from a nozzle and string,
3. Make sure the bed heats up before nozzles start, bed is much slower, otherwise , meantime the PVA oozes out of the nozzle significantly.
4. Ensure printer waits for the target temperature of a parked nozzle is reached before starting to move for printing. This may not be the default as you would expect as Raise3D does not list it's printer with a 0.2mm nozzle as a standard, so you have to create a custom one, the wait for bed to reach required temperature before heating nozzles has to be specifically set. Curiously the 5-step calibration firmware does support the 0.2mm, just don't expect any help from Raise3D beyond that.

Matte PLA

1. Rough surface creates much friction when transported from spool into the entry of the machine and within the feed tube to the extruder. Upgrading the feed tube from 2mm bore to 2.5mm bore helped for this direct drive system. In an indirect Bowden drive the much greater retraction distances required can cause the filament to zigzag in the tube as opposed to sliding within it. This increases the friction as the peaks of the zigzags dig into the tube walls.
2. Friction causes the layer being pulled off a spool to bury itself in the layer below, increasing friction enormously.
3. Spool and feed friction reduces the ability of the extruder to deliver the filament through a small bore nozzle.
4. Any obstruction to the outlet of the nozzle, including picking up stray filament from the other extruder. will cause a flow restriction and not a total blockage, but, the serrated extruder wheel will again grind itself into the filament, feed stops until reloaded. Just clipping off the first 10mm so that the ground part is below the serrated wheel and reloading suffices.

New Test Part

Having had little success using the readily available temperature and retraction towers it was time to create a specific one aimed at what I am trying to ultimately print that will yield results in a much shorter print time.

What I derived is shown below, it incorporates small diameter posts and an overhang that needs supporting above and beyond its base plate.

CAD Model



Continuing the search for the magic combination of spells to make this all work

So first I tried low nozzle temperatures to reduce PVA stringing, keeping flow rates low to compensate as the filament would only just be flowing and resistance to flow was already causing filament grind.

No worthwhile improvement.

Issues when using a nozzle different to the standard size of 0.4mm bore

Manufacturer temperature and flow rate data appears to be for 0.4mm nozzles.

Nozzle temperatures must be at the recommended lowest temperature or lower if possible to prevent material degradation which inhibits flow, except for the first layer to aid adhesion.

Bed temperatures need to be relatively high for the first layer otherwise the very small diameter filament that is laid will not stick, so it stays on the nozzle, degrades, blocks it, result, filament grind, so flow stops completely. Need to unload and reload past the ground notch in the filament. (If bad the serrated drive wheel needs cleaning too).

Bed temp minimum of 55C otherwise first layer does not stick to bed, stays around nozzle tip and blocks it, filament grind, all stops.

Printing at minimum temperatures with no or little cooling helps for a few layers, in the end given up using PVA, next try a BVOH which can print much like a PLA, e.g.

BASF Ultrafuse neutral BVOH filament

https://www.123-3d.co.uk/BASF-Ultrafuse-neutral-BVOH-filament-1-75mm-0-35kg-i7711-t26987.html

 

Well, started using the BVOH which at least strings less and is stated as being quicker to dissolve from the print but all is not well extruding the matte PLA, jam after jam, this was getting extremely tedious.

To deal with stringing more effectively I started using wipe walls, this surrounds the item being printed and is used to clean and prime the flow from a nozzle that has been brought into action to start its task for a layer.

The effect on the print was that after a few layers the matte PLA no longer extruded, however it could be unloaded and reloaded, eventually after some time with Google the cause was not what I had thought was, of course you can only take effective remedial action if you can identify the cause.

It turns out that the issue is what is referred to as heat creep, this is the unwanted transferrance of heat from the hot end up the filament towards the extruder.

By design it is imperative that while the nozzle is heated to allow the required flow of plastic the plastic filament must be rigid enough to be fed by the extruder. So above the nozzle there is a combination of parts to restrict heat flow upwards including a fan attached to a heat sink, the connection between the hot end and the heatsink being known as the heat break, i.e. it seperates the hot zone from the cold zone.

The upper cold zone parts insert and are clamped into the heatsink assembly, above which is the extruder drive wheels.

Image from Raise3D forum.



Image source:

https://forum.raise3d.com/viewtopic.php?f=20&t=21068&p=57966&hilit=ptfe#p57966

An in depth description of hot end operation can be read here :-

https://e3d-online.com/blogs/news/anatomy-of-a-hotend?

In the case of the E2 the heatsink is part of the head assembly and is ventilated by a fan.

The Issues with an E2

Cooling of the heatsink by a fan suffers from two aspects when the extruder/hot end is parked.

• The fan is always partly covered by a cosmetic housing.
• Fan vent is about 30% behind the built in spool boxes.

So even with the front door and lid open the heatsink fans have restricted airflow to them when parked.

So far I have resorted to not using the built in spool holders becuase

1. The PVA needs dispensing from a humidty limiting box, typical relative humidty here in the UK is 35 - 50%, preferable to be below 25%.
2. The matte filament cannot be dispensed from the internal spool holder due to friction of the filament off the spool and at an angle across a card spool side which creates too much friction. So again use a seperate spool holder dispenser outside of the printer enclsure.

So I do not use the two internal spool containers the following changes were made.

1. Removed the cosmetic covers from the extruder/hot-end assmblies
2. Removed the internal spool holders

Ventilation of the cold zone is greatly increased, clogging of the extruder with PLA has stopped.

An obvious question: Why was I able to print parts with a 0.2mm nozzle before without issue?

Quite simply the parts were printed with one filament from one extruder/hot end which kept on the move so that heat creep did not occur. Printing alternately with both extruders one sits parked while the other extruder prints. It is during the parked time that heat creep occurs, despite parking at 180C, well below the normal printing temperature.

Start Again: Printing Parameters

Created a copy of the high quality PLA template for Raise3D Premium PLA, 0.4mm nozzle.

Changed nozzle diameters from 0.2mm to 0.4mm

Material profile for PLA, changed flow rate to 33%*

Created a BVOH profile from the PLA profile also with 33%* flow rate

* Note: the 33% as opposed to the normal (for 0.4mm nozzles) of around 90% is a tip I found on a Raise3D forum post. So this time I keep the printing speeds the same but lowered the flow rate.
​

Print temperature held low at 200C, parked temperatures 180C.

​

Custom Test Piece Print Result

My custom test piece, PLA printed well, BVOH good but still a little stringing, no issues with clogging when a wipe wall surrounded the printed part, 3 prints in succession turned out this way.

This is by far the best print yet, very promising, at last.

Initial chassis section

This contains a pair of W-irons, part of the chassis framework and two pairs of brake shoes with hangers.

CAD Model





Result

As printed it looks quite a mess with the large octagonal wipe tower for cleaning ooze from a nozzle on return from parking and priming the flow.





The larger pieces of support material easily crumbled off and the remaining dissolved off in plain water within minutes, a light scrub with a small tooth brush to remove some remnants, probably this would have rinsed out anyway. The result is as shown next.





Positives

Detail better than expected, including the builders plate (admittedly only an outline, fine for a 1960s model), angle and channel chassis framing and suspension parts.

Negatives

Brake hangers were brittle and broke off easily, not really surprising as their cross-section is tiny.

Layer adhesion is weak as can be seen on the rear of the W irons, a pitted appearance rather than a smooth surface.

Lifting experienced in one corner, so bed adhesion needs improving.

During printing there was the occasional grinding sound from the support material extruder when about to start printing.

Summarising

Overall this was a successful trial print and a major step forward in the printing process, some three weeks of grief before being able to determine the problem was heat creep and so be able to eliminate it.

However, work to do :-

1. Raise the PLA print temperature to create sound layer adhesion.
2. Improve bed adhesion using JT16 bed adhesive
3. Further reduce stringing of the support BVOH material
4. Reduce the speed and distance of retraction of the BVOH materal to eliminate grinding
5. Expect to print the brake hangers with shoes as separate parts orientated on the bed for strength.

 

We'll see, so far they are too fragile, layer adhesion issues, but I am working on it

Jim

 

The next run of prints address points 1 to 4 above and in passing modify the bearings in the axle boxes to accept brass bearing cups.

Brass Axle Bearings

These are a necessity to give some running life to these wagons, the bearings of choice are Romford shouldered bearings which require a hole 2mm deep and 2mm diameter in the axle box, flange diameter 3mm, 0.3mm thick. Hornby 12.6mm 3 hole disc wheels are to be fitted.

Initial print showed that the flanges needed to be flush fitted into the back of the w-iron/axle box, so the second print as below.





The weakness in the w-iron must be resolved as springing the irons to insert the axles can cause them to fracture.

Improving w-iron strength.

These are only 1.4mm thick which does not give much scope to effect a solid print, to quicken up development of this issue a small test part was created which will print a lot faster, the w-iron prints take nearly 2 hours, of which 30mins or so is due to swapping extruders at nearly every layer.

This simple test part includes an approximate iron shape including an archway.

CAD Model



With the settings as for the w-iron test above this showed the same surface defects, primarily up to the arch, after that a good surface finish.

Adjustments that minimised these imperfections :-

• All layers set to have concentric ring infill, otherwise alternated between concentric infill and diagonal solid fill.
• The ordering of shells changed from inner, outer, fill to outer, inner, fill

There is still a little discontinuity on the sloping faces but this is acceptable, after all filament printed parts are not naturally smooth anyway.

Support Material

This was applied to the arch but was very reluctant to stay in place on the base plate of the print resulting in barely adequate support and worse strands across the PLA printed surfaces which can only weaken the print even when the visible part of it is dissolved away.

However the W-iron test part still has this pitted effect, it occurs where there is not a continuous flow due to gaps in the layer e.g. the open vertical slot in the W or when a support extrusion is required.

Not there yet - where next?

I need to investigate further whether this is a dampness problem, even a small amount may be enough, or perhaps it's just the colour of my socks today (as paul_l warned me a year ago ).

Tried with HatchBox PLA+ Pro and that was far worse, even when using basic PLA settings.

How dry can you get a spool of filament, I suspect domestically you can only really dry the outermost layer, but that is also the layer to get damp first, so if you only use one layer you should be okay.

What I was considering as a single or double layer depth spool a couple of posts or so ago when the matte filament was burying itself into lower layers when being pulled off the spool may be of use here.

That is, modify a spool such that it only takes a layer or two at a large diameter, i.e. increase the spool core diameter substantially to e.g. 180mm diameter (spool O.D. 200mm). That could contain ample filament for testing and probably a complete Conflat P chassis.

Performing the core diameter increase.

Add 6 or 8 bars equispaced at a suitable distance from the core, by doing it this way ample ventilation should be possible to both sides of the filament to be dried, this should allow me to confirm whether dampness really is the issue.

I have tools to join filament if necessary to avoid a lot of off cuts.

Right, time to print some shaped bars on the DaVinci.

 

Drying Filament on a Modified Spool

I have a transparent spool available which will be easier to modify when drilling the securing holes for the bars.

The bars are simply drawn and printed in 3 pairs.

CAD Model



Fitted to the spool.



The empty spool weight is 274g, with one layer of the Eryone matte black added, total weight 333g.

Dried for 4 hours at 50C as before, turning hourly, 180,90,180 degrees.

Weight at end of 4 hours, 333g, ho hum, but still it's only about 60g of filament.

Next. try printing with it to see if there is any change in its characteristics.

No difference.

However, it does prove that there is not a need to dry the filament in smaller batches.

If only the outer layer on a standard spool is the only one that really gets dried, it is of course the one that is most exposed to the atmospheric damp in the first place.

I know it flows well when creating the wipe tower sides, but when it is printing a complex cross-section it does not, so if it's not a dampness issue it must be another flow control issue, retraction settings are the most likely and also the most sensitive.

Sensitive is the watch word when using 0.2mm nozzles, small bore, and small differences in settings make big difference in results.

Time for a ponder, a restful word!

 

Soluble Support Materials

There are at least three scenarios where creating supports with the material used for the required object is not suitable.

The printed part

1. is delicate and will not withstand the breaking off of the nearby supporting material.
2. Has inaccessible supports when the item has been printed.
3. Is small and leaving a 1 layer gap for the next layer to droop onto (the default for filament based printing) is significantly large, e.g. a 0.1mm droop on details of even 0.4mm is noticeably bad, but inconsequential on something 10mm deep.

My Conflat chassis has much of (1), some of (2) and plenty of (3).

The two least expensive soluble materials are soluble in water, the more expensive requires a special bath too.

PVA

The cheapest option.

The big negative with this was that as soon as it is nearly hot enough to print it oozes out of the nozzle far too easily, then doesn't stick to a wipe tower or wall, strings over everything and also gets on the other nozzle which is printing the required part in PLA.

Initially this contributed to nozzle jams, which were eventually traced to heat creep as discussed in an earlier post on this thread.

BVOH

BASF moderate price, Xioneer very expensive.

The BASF product (BASF Ultrafuse neutral BVOH filament) also suffered from significant oozing, just like the PLA, but it certainly dissolves readily.

The Oozing Scenario

Both oozed significantly during the switchover between the two extruders, even though parked at below melting point temperature, they ooze when they

1. finish their layer whilst the nozzle cools to below their melting point
2. are pre-heated to start printing as soon as possible when it is their turn

What about retraction? Shouldn't that fix the issue, well, not always.

Retraction only takes the pressure off the molten contents within the nozzle but these two soluble materials will almost run out of the nozzle unaided, or worse curl upwards on the nozzle where it becomes impervious to wiping.

Eventually it builds up around the nozzle and gets dropped where you don't want it like short pubic hairs which are blown everywhere by the extruder job cooling fan, or contaminate the PLA nozzle.

After extensive testing with the enormous amount of interrelated parameters driving me to wonder why on earth I even thought about 3D printers in the first place I have finally found a combination of parameters that appear to deliver the goods, on a good day of course!

So I will just summarise my results with BVOH rather than bore you into a coma, the PLA is well behaved.

Nozzle Temperature

I have found that a very narrow window achieves almost string and ooze free printing of BVOH, the recommended print temperature is between 190 and 210C, melting point 175C.

Generally material specifications are based on using 0.4mm nozzles, so when using a 0.2mm nozzle the cross-sectional area of the opening is a quarter of that for a 0.4mm nozzle and the data sheet for this BASF material is no exception, they specify >= 0.4mm.

For a given speed across the print bed the flow rate is therefore lower*, hence the time spent by the plastic material being heated is longer and so heats up much quicker and higher. Typically nozzle temperatures can be lowered for nozzles of less than 0.4mm bore.

* Note: base material flow rates may need to be lowered too.​

When parked the BVOH nozzle temperature is set to 170C, but it takes a few seconds to drop below melting (oozing) temperature which is when most oozing occurs, typically a strand of 5 to 8mm long.

For the BVOH I lowered the extrusion temperature to 182C, barely above the melting temperature of 175C, 184C+ for oozing to become a nuisance. Now take into account the calibration of the heat sensor on the hot-end and the hysteresis of the temperature control, the nozzle temperature can vary by an indicated +/- 2C, this shows what a narrow band allows printing without nuisance level oozing.

This may seem ridiculously tight figures but it works on my E2, probably different on another E2.

When priming there is a couple of extruder grinding clicks when it does a poop at the end of the priming strip but otherwise no grinding at other times, to avoid this the temperature needs to go up, and ooze we go again!

Layer adhesion is weak but not an issue as the material will be dissolved away, very good bed adhesion is achieved using JT16 adhesive on a AJOYIB bed and a support solid fill first layer.

Slicer Issue - Support Horizontal Expansion

Using the latest version of IdeaMaker (v5.2.3 at this time, May 2025) it has some annoying little quirks and the one present on this job is where you set the amount of X-Y distance the support expands under an item, it works erratically on small parts, even varies on whether a part is aligned long side to X or Y or even 45 degrees to both.

On one part of the solebar which is a typical U channel it refuses to insert the lower layers of the support in one place unless it is increased from 0.2mmto 0.45mm, which adds over 30 minutes to the print time by being excessively larger than necessary for probably 90% of the print job. Without it the job fails.

On the earlier w-iron tests only 0.36mm was required with rectilinear support pattern.

Support Pattern

I have found the Gyroid pattern seems to work best for this job as it is printed mainly as continuous waves, which means less retractions.

Wipe Tower/Wall

Due to the poor layer adhesion of the BVOH printed at such a low temperature, printing separate towers to minimise the PLA nozzle picking up strands of BVOH was a failure because the BVOH tower would break free from the print bed, chaos then begins as the broken tower gets dragged around.

However using nested wipe towers, in this case the BVOH is printed within a PLA tower stops the BVOH part from breaking away, this only works now that the oozing and contamination of the PLA nozzle has been almost completely eliminated.

Within a separate tower any stringing stays within it, if a wall is used then the strings passes over the printed part, just where you don't want it.

With IdeaMaker I use a small octagonal tower that is 10mm wide, and each nozzle makes two loops which is adequate.

And, on the other serious issue. heat creep.

Heat Creep

Having identified that and enhanced the cooling of the hot-ends not one jam or clog has occurred with the PLA extruder since then, before it was occurring within minutes, bad when considering the need for 5 to 7 hour print jobs.

 

Correction to the above header has been made, "Slicer Issue - Support Horizontal Expansion" was "Slicer Issue - X-Y Offset", not the same.

 

A Rolling Chassis

I have now taken the CAD model of the Conflat P and 'dismantled' it for printing in sections compatible with 3D filament printing, this unveiled a few more unstable parts of the chassis model, e.g. parts which when edited would not reconnect due to the original transformation from an STL file. Enough said about that earlier in this thread.

As each part was separated from the chassis a means of registering the parts back on to the chassis were added, pins, holes, tongues etc so it could be re-assembled as a simple plastic kit.

Anyway, time to start printing the components just to see what I am up against using the settings derived from testing just part of the chassis. The most crucial is the basic chassis frame, reduced to being a flat component on its upper face and braking components removed from its underside.

CAD Model




As printed with the BVOH support filament showing as white.



Cleaned up with wheels fitted, the channel and angle cross sections of the chassis frame are actually printable, although attention is required to finer parts such as the ends of the buffer beams.

Matte PLA does not seem to have the same layer adhesion as non-matte PLA, I may need to revert to non-matte for detail and added toughness. Brittleness is quite noticeable on the tie bars across the w-irons below the axle boxes. Easy to break when fitting the brass bearing cups.





It does not bear close scrutiny as the surface finish needs a fair bit of improvement which I believe is part PLA temperature and part incursion of the support medium (BVOH) against the outer surface of the PLA solebar and W-irons.

Using a very fine gyroid support structure has shown up clearly that a gap is required between the vertical face of the BVOH and the PLA as its pattern shows up on those parts.

This was not so obvious with the previously used rectilinear support pattern.

Overall the chassis frame appears to be strong enough to be practical, it just, (just he says) mainly needs cosmetic improvements. It stands well on its wheels, there is no twist or distortion apparent which is also a promising indication that this could become a functional wagon.

Fitted with wheels I can now assess the mounting points for couplings and which buffers to use, white metal or a mix of printed bodies and metal buffer heads.

These wagons would be run in block trains and not split up in transit so they would not stop at my marshalling yard for splitting. So, a simple coupling could be fitted between the wagons within the block. At each end of the block, I would fit a Dublo type to run with my CoBos and Stove brake vans.

A mount compatible with both is desirable, and the couplings printed as required, I have already been able to print a working version of the larger, and my preferred Dublo/Wrenn coupling in PLA.

I could, if daft enough mount a NEM socket to cater for all eventualities, such as a bar which clips into a NEM socket, a plain version similar to that created for the Airfix 14xx DCC conversion where I semi-permanently coupled the loco and its auto-coach.

 

Couplings and Chassis Prints to Test Run

Another alternative for the couplings using the KISS principle is to use British Trix wagon couplings which on first viewing are just the right height when attached directly to the underside of the chassis.

These couplings are slim, and quite effective at coupling providing they have a little support where they come out from under the buffer beam. This can be achieved by a letterbox slot as shown below.

Various options for retaining the couplings were tried out

1. Screw into a raised boss on the underframe as on Trix wagons, but the boss wall is too thin to be effective and snaps off the underframe easily.
2. No screw, just the boss, I found that when fitting the couplings for use with a screw they tended to stay on the boss when pushed and pulled by hand, really simple but how long effective. These bosses would be slightly higher than those for a screw and printed solid.
3. Shouldered pin that push/clicks into place like filament printed pins like to push fit into holes, the ribbing from the layered printing process makes this possible.

The third option is the neatest and simplest and will be tried in practice first.

Thickening of the buffer beam between the solebars was necessary for strength, the outer ends kept as original for appearance.

Printing in matte white as it is much easier to examine the resulting prints than with matte black.

CAD Model - Test




Test Prints - Part with Buffers




CAD Model - Basic Chassis + Pins



Couplings fitted to a Basic Chassis







The pins are printed four at a time with a chassis to allow for any defective prints and spares as they can't be removed again in one piece.