machines which I have made, am making, or intend to make, and some other stuff. If you find this site interesting, please leave a comment. I read every comment and respond to most. n.b. There is a list of my first 800 posts in my post of 17 June 2021, titled "800 Posts"
It has been cold here during the current lockdown. And I mean temperatures. Not by American midwest standards by any means, but since we are confined to our homes except for limited predefined purposes, some days and nights are chilly. Down to 5-8ºc here.
I have been spending a lot of lockdown time doing 3D prints. And really struggling to get decent results.
Some of my GSMEE colleagues have been urging me to make an enclosure for my 3D printer. To be honest, Stuart T had urged me originally to buy a printer with an enclosure, but I pressed ahead and purchased an open structure model because I wanted the extra print size it offered. The Creality CR10s can print up to 300x300x400mm which I have fully used for my Ottoman bombard prints.
But in recent weeks, with the onset of the cold weather, I have noticed a distinct deterioration in print quality, particularly with poor layer adhesion when printing overnight, when the house heating is turned down or off.
So I decided to make an enclosure!
But, I did not have the materials on hand, and visiting hardware stores is verbotten with lockdown rules.
So, don’t laugh. This is what I cobbled together……
A couple of cardboard cartons, an artist’s A0 paper case (SWMBO hasn’t noticed it missing yet), and a blanket.
The heated printer bed is the heat source, at 50ºc. And I was surprised at the temperatures reached inside the rickety construction.
The steep temperature rise on the left is inside the enclosure after printing started. As you can see, the temperature rose from about 18ºc (room temp), fairly quickly to over 30ºc. After midnight, when the house heating was turned off there was a slow drop to 25ºc, and then a further drop to 18ºc when the printing finished and the bed self turned off.
The temperatures were measured with this gadget. A temperature/humidity logger.
Inkbird Temperature/humidity logger.
And the printing result??
The printer inside its enclosure, on the dining table.
This is the best quality print which I have had since the onset of winter weather. It is solid, water tight, and a reasonable finish. 0.2mm layer height. It is a molten metal pouring funnel, so I was not trying to get a super smooth finish, just an intact water tight object.
As soon as I can get access to Bunnings, I will make a more purposeful enclosure. Meanwhile, the cartons and blankets can remain in use.
I have been attempting to print a 1:10 scale barrel of the Ottoman bombard, in PLA, so I can make a cast in jeweller’s investment, and use that to pour a bronze version of the cannon.
I borrowed a big furnace to melt the bronze, and broke 2 ribs unloading it from my vehicle. That was about a month ago. They still ache a bit, but apart from careful positioning in bed, are steadily mending. I have to sleep on my back, which would normally make my snoring unbearable, but the CPAP machine is working quite well. SWMBO absolutely insists that it is in constant use.
And I have purchased a length of 5″ stainless steel pipe to make the mold.
I have featured the Ottoman bombard in previous posts, having made a wooden version some years ago. It is over 500mm long, and 107mm diameter. In 2 pieces with a big thread joining the pieces.
Just to remind you of the appearance of the bombard. This is the wooden version. 500+mm long, 60mm bore.
I can’t really justify a bronze version. It will weigh close to 20kg. But it is a challenge. And I think that it will look more authentic in unpainted bronze.
I printed the breech part a few months ago.
It is 240mm long, and with some post printing finishing will come up fairly well. The thread will be replaced by a redesigned thread. The original male thread on the Royal Armories bombard was tapered, so I have printed a tapered PLA version and will cut off the thread pictured above and glue on the new one before casting. I have tested the tapered thread in a test piece of printed barrel with female thread (which is not tapered) and it does go on much more easily than the parallel version, so that gun maker (Orban, the Hungarian or German) knew a thing or two.
3D PRINTING THE BARREL.
The barrel is 315mm long. My printer has a maximum print size of 300x300x400 mm so I was not anticipating any problems. I knew from the slicer program that it would take 2/3 of a 1kg reel of PLA, so I bought some new transparent PLA, thinking that it might melt/vapourise more completely in the burn out cycle of the production than the coloured PLA.
So I tried to print it. I have lost count of the number of unsuccessful attempts. Each time the print would start well, but at some point, sometimes after a whole day or 2 of printing, the print would come loose from the printer base and I would have clean up the mess of PLA spaghetti, and start again. I cleaned the printer base thoroughly. Scraped it. Wiped with acetone. Re-levelled it multiple times. But every time the print would break free.
I also noticed that I was getting a lot of stringing, and lumps of PLA would form on the printed surface, cool and harden, and sometimes the print nozzle would hit the hard lumps. That is when the print would loosen from the base and eventually break free.
I have been using a 3M printing cover over the aluminium printer base, quite successfully for over a year. Maybe the cover had worn out. I looked up the P.I. for the cover, and yes, it has a stated expected life of 10 uses! So that was likely the cause of the adherence problem because I must have used that cover at least 50 times!
I had no replacement 3M cover, so I reverted to the original cover supplied with the machine, which was boro-silicate glass. Initially it worked well, with good print adherence, but the hard lumps were still forming, and when the nozzle hit them, there was enough force to break the glass plate free.
What could be causing the hard lumps?
I watched multiple YouTube videos. Re-levelled the bed again. Checked every nut and bolt on the printer for tightness with no problem found. Checked the Z axis for level.
By this stage I was contemplating buying a new printer. Maybe one of those liquid + UV light jobs. But one of those big enough to make my barrel would cost thousands. So I got a quote from a professional printing service to print the barrel…. almost $AUD600. I would have done that, but the print is destroyed in the making of the cast, and it is possible that more than one attempt of bronze casting will be required. I was considering abandoning the entire project.
One last try at a print. I replaced the 3M cover with a new cover, and started a new print with a new reel of red PLA.
All seemed to be going well.
The print was adhering solidly to the new 3m cover. The hard lumps were still appearing, but the print head ploughed through them or knocked them off completely. The problem was, that after 3 days of printing, with 10% of the barrel still to go, the multiple jarrings were producing axis shifts. The appearance was pretty bad, but I figured that I could fix it with some extensive post printing hand finishing.
By this stage the print was almost 300mm high, and I could watch the laying of the PLA extrusion from the print head directly. In retrospect I should have used a mirror to do this at a much earlier stage.
What I saw explained the issue of the hard lumps appearing.
PLA was slowly oozing from around the base of the extruder nozzle. It was gradually building up into a pea size lump, and eventually falling off onto the print face!
So, I paused the print, picked off the accumulating lump, and watched some more. The same thing happened.
Why was the base of the nozzle leaking? Another pause. Checked the tightness of the nozzle. It was totally loose. About a full turn!
Tightened it up. Resumed printing.
The next layer did not adhere at all to the previous one, because tightening the nozzle had lifted it at least one mm.
The almost completely printed barrel. Lots of stringing. No hard lumps in this picture. This is in the dining room of my house. The room has been unused since the start of Covid. Quite handy and warm for printing.
I thought that I could start a new print of the final 10% of the barrel, and glue it to the part pictured, but when I examined it, the layers were poorly adherent, and falling apart. It went into the plastics bin. I expect that the loose nozzle caused multiple print faults in x, y, and z axes.
A record of printing failures.
So, I am now 32 hours into the next attempt, with 47% completed.
See the difference? No stringing. No lumps. Quite a reasonable surface. Fingers crossed.
I have been tearing out chunks of hair, and gnashing teeth because prints are failing due to loss of adhesion. And the problem has progressively worsened to the point that the failure rate is now 100%.
And I am thoroughly fed up with throwing bird’s nest conglomerations of PLA in the bin.
I have tried the following remedies…….
Varying the bed temperatures and extruder temperatures up and down. Various combinations and permutations.
Cleaning the bed after every print, with acetone and scraping.
Checking and rechecking the bed for level.
Replacing the extruder nozzle.
Changing the PLA to brand new stock.
Printing on a different part of the bed rather than the default centre.
Today, I searched the net for solutions, and I decided that I had tried all of them, EXCEPT, replacing the bed surface.
I have been using a 3M product, designed for 3D printing, designated 9080A. Today I RTFM. If you do not know what that stands for, try “Read the F’ing Manual”. So I RTFM’d.
It seems that 9080A is good only for 10 uses, then should be replaced. OK. Penny drop time. I have been using the printer for 18 months with the same 9080A sheet. Maybe 50 times?
Back to Amazon, order some more 9080A sheets. Meanwhile I will revert to the borosilicate glass sheet that was originally supplied with the printer.
Almost finished the model Armstrong 80pr RML, and just starting another project. I have mentioned it in previous posts…. a 1:10 scale model of the 17 ton Turkish bombard, which currently resides at the Royal Armories Museum, Fort Nelson, Portsmouth, UK.
The original was in 2 pieces, to make the casting process manageable, and presumably to make transporting the monster cannon more manageable. The museum states that another reason for the screw thread join of the 2 massive parts was to separate the halves for reloading, but I can find no substantiating references for that statement. And it does not make sense to my conception of what would have been involved in the reloading process.
At 1:10 scale the model will be over 500mm long, and will presumably weigh approximately 17kg (37.5lb). Each piece will weigh 8-9kg. I will make the model in 2 pieces, for authenticity, and to make the casting more manageable, and to make the 3D printing possible. My 3D printer has a maximum model size of 300x300x400mm.
I spent several days drawing up the breech and saving it as an stl file, for the slicer (Simplify 3D) to process. The slicer predicted that the print would take 51 hours, and consume 697g (1.5lb) of PLA. I used 0.2mm layers, with 8 top, 8 bottom, and 6 side layers, and 10% fill, and since there wee some 90º overhangs, I decided to add supports.
And guess what. The print took 51 hours, and consumed most of a 1kg roll of PLA.
I chose to operate the extruder a bit hotter than normal, at 225ºc, and heated the platen to 65ºc. I wanted to make sure that this print was water tight for the moulding process, and remained adherent to the platen for the duration of the print. I accepted that the detail of the print surface would be a little coarser than could be achieved at a finer layer thickness, but the benefit would be increased water tightness.
The Ottoman Bombard at Fort Nelson. In the background is the barrel for the supergun which Saddam Hussein ordered, but was prevented from being exported from the UK.After about a day of printing. On our dining room table (which I made many years ago).Phew! Printing completed.Most of what can be seen here are the supports.It took about an hour to remove the supports. They were particularly resistant to remove from behind the pins.I will spend a few more hours sanding and filing and filling the surfaces, before making the molds with the investment powder.
I am still drawing up the barrel. Well, actually, it is fully drawn up, but I am refining the drawing of the Arabic script which is embossed on the muzzle. It is quite difficult to convert the squiggles and patterns to vectors, which can be used to produce the STL file for the 3D printer.
The Arabic patterns and script on the muzzle. At 12, 4 and 8 are floral patterns. The calligraphy reads “Help O Allah. The Sultan Mohammed Khan son of Murad. The work of Kamina Ali in the month of Rejeb. In the year 868″. (CE. 1464). p.s. I did not previously notice the alien watching me , top right.
The white pieces were printed several weeks ago, then the black pawns. But I had 2 failed runs when printing the black major pieces. The failures seemed to be caused by failed adhesion of the pieces to the platform. In each case, the runs were progressing nicely, but failed after about 20 hours, in the middle of the night, covering the 7/8th completed pieces with PLA spaghetti.
The settings were exactly the same as the white pieces, so why the sudden failures? Is the black PLA different in some way?
So I asked my colleagues at the GSMEE. (Geelong Society of Model and Experimental Engineers), some of whom are experienced 3d printers. It was suggested that perhaps I had turned on the cooling fan too soon, after layer 1. So I changed the setting so the fan did not come on until after the platform and one layer of the pieces had been completed. And the result was excellent! See the photo.T
The black pieces, after a quick clean up. A successful run, which went for 26 hours.
Of course the colour of the PLA is irrelevant. The PLA will be melted, vapourised and burned out after the molds are made. But I could not resist the opportunity for a photograph.
Next, to make the wax and PLA trees, and make the molds.
Still thinking about what metals to choose, and how to colour them. The pieces could be used just as they are, but I really want to feel the weight of real metal pieces.
And although I claim that the 3d printing is complete, the assumption is that there will be no casting failures. I could well be printing more pieces.
Each piece takes 2.5 – 4 hours to print at the high resolution which I require to produce a good finish. The printed pieces will be attached to a tree, then encased in jeweller’s casting medium inside a steel cylinder. When set, the cylinder is heated to 200-300ºc to melt and vaporise the PLA, producing a cavity in the casting medium, into which the molten metal will be poured. The mould is baked for about 6 hours to thoroughly dry and harden it before the metal is poured into it.
There are 16 pieces in each army of a chess set. So 64 hours of printing for each colour. Plus failures. So far, in about 5 days of printing, I have produced the whites. That has taken almost 1kg of PLA, one roll. PLA is not expensive. I paid about $AUD22 per roll, including postage. Lately prices have risen to around $AUD30 per roll.
These are examples of a print run failure. This run was almost completed after 24 hours, when for some reason it just stopped. It was overnight, possibly a short power outage. Another run failed due to poor plate adhesion, again near the end of a run. I solved that issue by turning up the temperature of the extruder to 220ºc and the temperature of first few layers of the platform to 70ºc.An army of pawns. One spare.Half ready for casting. Now printing the opposition (in black PLA, only because that is what I have on hand, plus it might be another photo opportunity.). Can’t wait to see these in aluminium and bronze.
There are 3 major components of each wheel assembly, plus the wheel, axle, and king pin.
The wheels, axles and king pins are straight forward metal turning, but the other 3, the wheel bracket, the king pin post, and the chassis bracket, are castings in the original.
For my 1:10 model I am planning to cast the king pin column, and the wheel bracket. But I will fabricate the chassis brackets.
There is one chassis bracket for each of the 4 chassis wheels, and they are all different. Front different from rear, left and right hand versions. And each one has angles of 90º, 30º, 20º, 6º, 2º so the machining was quite a mental exercise. No major stuff ups though.
Here is the main component of the left hand rear chassis bracket, being held in position. It will be bolted on later, and have several flanges silver soldered to it. Those M2 cap screws will be replaced by rivets eventually.
Meanwhile, having decided to cast the king pin casing, and the wheel bracket, I spent many pleasant hours (or was it days?), drawing them. Then yesterday, I 3D printed an example of the king pin casings.
2.5 hours to print PLA examples of rear (left) and front king pin casings. I need to see the original cannon to check some details before committing to cast these in bronze. The PLA parts will disappear during during the casting process. (A pity. They are quite attractive No?) You can see why I chose not to machine them out of bar stock. 3 pin holes in the left hand print ? the result of not storing the PLA spool in a dehumidified container.
So, it might not look like several days of computer and workshop time, but that is how long it has taken.
In Australia we have had some easing of Covid-19 restrictions, but not opening of museums or historic collections of cannons. So I still cannot go to Warnambool (a 2.5 hour drive) to check details on their Armstrong 80 pounder rifled muzzle loader. Flagstaff Hill Maritime Museum does not answer their phone. Hmmm. Maybe I could climb the fence and sneak in…… but maybe not.
My model Armstrong cannon has some components which will be difficult to machine, and would involve silver soldering many tiny pieces.
For example, the steel brackets in which the wheels are supported, and the centre column.
There are 4 trolleys like this. Each one has 2 or 3 wheels. It is a Z shaped profile with 3 gussets visible and 2 more inside.
The centre column. It could be fabricated.
But being basically lazy and always looking for the easy way out, I have decided to investigate the possibility of casting these parts. And some others.
So I have printed them in PLA filament, with a view to a “lost wax” type of casting process. It will be “lost PLA” of course. Maybe doing the casting myself. But also checking the possibility of having it done professionally.
The PLA printed parts which will be melted and burned away in the casting process, have to be as well finished as possible. So I have been experimenting with various settings in 3D printing. One problem is that the molten plastic thread has to be supported. Overhangs up to 45º or even 60º can self support. And even horizontal overhangs can self support if the gap is not too big.
But this gap, about 20mm, proved to be too big…
The threads are partly bridging the gap…
Horrible. It is the underside, but even out of sight, it is unusable.
So, I am printing up some supported versions, even as I type this. And I am going to look at some casting equipment which I might be able to borrow. Apparently the gas furnace is very noisy, and it needs a home with no close neighbours. List…. a furnace capable of melting bronze, a crucible, investment casting powder, protective gloves, helmet or face mask, leather apron, tongs, slag ladle, a casting box. There are many YouTube videos on the subject of lost PLA casting. Watch this space. But if the quote for professional casting from my printed molds is not too fierce, I will probably take that path.
johnsmachines 