Making Mendel90

This beautiful curved model by virtox had proved extremely tricky to print for various reasons. My best attempt had been in black ABS and although the print completed successfully (unlike several earlier attempts) it cracked in several places whilst it was still printing. This, more than any other, was the model that prompted me to build a heated chamber for the Mendel90.

Here is the print, lying on its side, showing the cracks that appeared as the plastic cooled and shrank causing delamination.

This copy was printed in natural ABS inside the heated chamber at 35 C and, as you can see in the picture, the surface is smooth and blemish free.

And here are the two prints side by side.

Last weekend, one of my sons gave me an unusual board game for my birthday (thanks, Luke!). The game comes from Germany and is called ‘Das Waldschattenspiel’, or ‘Shadows in the Woods’.

The contents include a large ‘forest floor’ mat, 10 wooden trees in various sizes, a die, a tea-light and 7 wooden pegs to represent dwarves. The simple rules involve the dwarves playing hide and seek in the woods by the light of the candle which is moved around the board by one of the players. I was intrigued by the concept but somewhat less excited by the suggestion of using a felt pen to draw faces on the men and then gluing on cotton wool and felt cones (included) to represent beards and hats.

It took only milli-seconds for me to realise that I could improve the game by exchanging the wooden pegs for proper dwarf models made on my 3D printer. 

A quick search on Thingiverse revealed that several models had already been uploaded but one stood out as being perfect for the job. He is a lovely little fellow called Gnome baldbeardington. It was a simple matter to resize the model down to 3 cm tall, the same size as the game pieces.

I haven’t played the game yet but already I can see future possibilities for adding levels of difficulty by printing dwarves in different sizes to make hiding easier or more difficult - and why  should all the characters be the same size? The potential for game modifications is just mind-boggling.

Most RepRaps are capable of printing whilst disconnected from the host computer as long as they have an SD memory card attached. But usually this means that the printer just carries on regardless until either it finishes the print or you pull the plug.

Not any more.

This clever little accessory is a real bargain at £30 as it includes an SD card reader, a click controller and a backlit LCD display panel. The display can show 4 lines of 20 characters and the click controller, in combination with the Marlin firmware that I’m already using, operates a menu system that provides extensive control over most of the printer’s functions.

You have to make your own case and control knob but , hey, that’s what 3D printers do best!

Two weeks ago, the heated printed bed on the Mendel90 suddenly stopped working. The cause was pretty obvious on closer inspection. The power connector had overheated and melted the plug.

Nophead had warned in his build instructions that it was better to bypass the plug and attach the wires directly to the metal body of the power controller but I didn’t bother to make the change and the plug failed :(

Worse still, when I tried to repair the connection using nophead’s method it still didn’t work, leading me to conclude that I had damaged the circuit board as well.

And then, in my sorry, tired state, I committed an even bigger blunder whilst testing the circuit, accidentally shorting two pins together and sending 12 volts down the 5 volt line which destroyed the processor and all four motor controllers.

Fortunately, I was able to afford a replacement board, swiftly supplied by Think3DPrint3D, and a set of spare chips to repair the old one. So now, two weeks later, I have a complete spare set of electronics which I intend to keep at hand for the Mini-Maker Faire in Manchester at the end of the month. 

And then after that, maybe they’ll find their way into my next 3D printer! Who knows?

Last week I entered the RepRap 3D printing competition (purely for fun) and came first in the judging. This was my entry.

Last night I finally solved a problem that has been dogging my Huxley printer for weeks. I previously wrote about how I had managed to find the correct PID settings for the Mendel90 and how this was now benefiting from very stable temperature control.

Well, I’ve been trying to achieve the same thing with the Huxley and failing miserably. All of my recent prints have had to endure ridiculous swings of +/- 20 degrees C and the quality has been suffering as a result.

I’ve read more and more articles and forum posts about the art of PID tuning and failed to understand why I couldn’t fix my problem. Finally, I stumbled across one line that proved to be the key to finding a solution - “sort out any hardware problems before attempting PID tuning”.

I took the hot end off the printer and inspected it carefully. Sure enough, there was a completely dry, loose connection between the thermistor and the heater block. No wonder the controller was having a hard time getting the temperature right - the readings must have been changing every time the head moved!

Squirting a little heat sink compound down the thermistor socket and replacing it cured all the temperature swings and now I’m looking forward to seeing a real improvement in surface quality of my prints.

I really wanted to use these gorgeous gears that I’d printed on the Huxley.

Herringbone gears are more accurate and less noisy than standard gears. They wouldn’t go on the Huxley because the hole is designed for an 8mm shaft and the Huxley’s is 6mm. So I was planning to fit them to the Mendel90 - but there was a problem. The large gear was taller than the block it was meant to fit into, so I designed a thin riser to fit under the block and lift it just high enough to allow the gears to turn.

Here it is in place onto the X carriage. I made it extend beyond the front edge because I’m thinking of mounting a fan duct to it.

And now with the extruder block on top of it. You can see the gears fitted now.

And here’s the view from the back.

Looking great and working perfectly. The only other things that have to be changed are the Z-height calibration (because the nozzle is 2mm higher now) and the extruder calibration (because a larger cog sends the plastic in at a slightly slower rate).

With the old cog it was taking 640 steps to extrude 1 mm of filament and with the new one it takes 698 motor steps.

One interesting complication which I hadn’t anticipated is that you can’t easily mesh and unmesh the two gears any more - the herringbone pattern prevents any sideways movement. Fortunately, the bolt can be extracted from the large cog without too much difficulty so I can still clean it whenever I need to.

This weekend I devoted several hours to that unavoidable ritual - ‘calibrating the printer’. Endlessly making minor adjustments to multiple parameters and printing out test blocks may not seem much like fun, but it can make a huge difference to the print quality and is really a necessary evil.

Instructing the printer to feed in 30 mm of filament only resulted in 25 mm going through so that was the first thing to correct.

Then I concentrated on getting the bottom layer to print nicely so that it was neither too squashed, nor too gappy.

I measured the filament diameter at multiple points (average value of 3.05 mm was very close to the expected 3) and entered all the values into the slicing software correctly.

Then I set the printer off printing solid blocks, expecting to see perfectly filled shapes of accurate size. What I got was that horrible mess at bottom left, with far too much plastic coming through the nozzle.

Once again I had to resort to the ‘extrusion multiplier’ (aka fudge factor) to get half decent prints. This value is normally set close to 1 but on my machine I ended up using a figure of 0.85 to get the prints as smooth on the top layer as they were on the bottom.

This isn’t a huge problem (I had the same issue with the Huxley and that prints really nicely now) but it is somewhat frustrating that having made such careful measurements, the values then have to be bodged by quite a significant factor to get decent results.

I finished up by having another go at the lovely tree frog model using my new settings and the tricky slope on the back is now better than it was, though still not as good as I would like it (new on left, original on right).

Isn’t life complicated at times.

The nozzle for the Mendel90 has finally arrived but I couldn’t work out how to fit it into the bottom of the extruder. After a quick email exchange with nophead I discovered that the STL file that I printed out is for a 16 mm hot end, but the one that I had bought (as recommended in the instructions) has a 10 mm fitting.

No big deal; it just means I need to print out a new base plate for the extruder. Of course, I can’t print it on the Mendel yet because it isn’t ready, so it will have to be done on the Huxley. But the Huxley is now fitted out with a new tiny nozzle for printing fine detail parts, like this lovely little frog.

I was half way through building a second hot end for the Huxley which will use the original, larger nozzle, and that’s the one I need to use to print the new base plate for the Mendel. With me so far?

So after work today, I finished building the second Huxley hot end only to find that one of the parts is too big to fit in the nozzle holder on the Huxley. So it’s back to the computer to make some modifications to the nozzle holder, then print out a new version to mount on the Huxley, so that I can then fit the larger nozzle back in order to print out the nozzle holder for the Mendel!

OK, it’s just finished printing. Here you can see how the middle channel gets wider at the far end.

Let’s hope this works.

The extruder needs a hobbed bolt (an ordinary bolt which has grooved teeth cut into it) to pull the filament through the heated nozzle. You can buy these for about £6 but there are instructions on the web for making your own so I thought I’d have a go.

You need an electric drill (preferably with variable speed control), an M4 or M5 tap and a jig made on a 3D printer fitted with roller bearings.

First I had to fit the bolt in the extruder and use a pen to mark the centre line where the hobbing needed to be done. I then use a scrap of tin, wrapped around the threads of the bolt to protect them in the chuck of the drill.

I used a half round file to make a groove in the correct place on the bolt. This helps the tap to cut the teeth without sliding all over the place.

Next, I fitted the bolt into the bearings into my printed jig (downloaded from Thingiverse) and screwed them down to a scrap block on the workbench.

I added a vertical screw to help me stabilise the neck of the tap bit in the right spot on the bolt. The first time I tried it, the teeth on the tap started wearing away, leaving barely a mark on the bolt. 

At this point I realised that my cheap tap and die set were not made of hardened steel and I needed something better.

A quick search on eBay turned up this one for only £3.50 and a couple of days later I tried again with much more success.

The really clever part, is that as you press the bit down, the bolt revolves by itself, cutting the grooves into a neat channel all the way round.

For a first attempt, I’m pretty pleased with the result and it seems to work well. Now I just need the heated nozzle to arrive.

On Monday this week, I took both of my Reprap machines, the Huxley and the Mendel90, to Sheffield for the monthly meeting of reprap users and builders. It was a great opportunity to meet nophead, the designer of the Mendel90, and ask him some questions about the next phase of building - the hot end parts and the wiring.

The Huxley performed well and I managed to print the large gear for the Mendel90 extruder at the meeting, which generated a good deal of interest.

As soon as I had finished making my eMaker Huxley 3D printer back in September 2011 I knew that, at some point, I would want to make a bigger and better printer. I just didn’t realise how soon that would happen.

Not that there’s anything wrong with my Huxley. In fact, quite the opposite. You see, I’ll be using the Huxley to print my next 3D printer, which is quite mind-boggling if you stop and think about it.

I have reviewed a number of different models that have appeared over the last 6 months but none has really caught my imagination quite as much as the Mendel90. The designer, nophead, began teasing us with posts on his blog in December last year but the finalised designs didn’t appear on Thingiverse until late February 2012.

There are probably several reasons I chose this design over many others I could have gone for, but the main one is that the whole project is so well executed. Every printed part, every nut and bolt has been accurately modeled in a mathematical 3D programming language called SCAD - a huge amount of effort - and then made freely available for everyone to use.

The next reason for choosing the Mendel90 is that nophead is undoubtedly an expert at what he does, as can be seen from the blog-posts going back several years, cataloging his progress in the art of printing solid objects on home-made printers.

Anyway, enough of the waffle; on with the story. On 20th February 2012, the same day that the plans were published, I printed the first piece of my new printer. Here it is - it’s the tray that holds the print nozzle, not that that matters. The important thing was that I was making a commitment to make this my next project.