Filament Splicer

Home Model Engine Machinist Forum

Help Support Home Model Engine Machinist Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.


Well-Known Member
Project of the Month Winner
Aug 16, 2013
Reaction score
We’ve been in “social isolation” for going on a month and all the interesting stores are basically closed. It’s also been unseasonably cold and we still have about a foot of snow in the yard so I needed a project.

The other day I was printing a new enclosure for my engraver and this was going to be a 4.5 hour print so I set a timer and left the printer to play by itself. When I went down to check on it 4 hours later I saw to my horror that I only had about 2 loops of filament on the spool. I’ve spliced filament a couple of times before using a short length of PTFE tube and a BBQ lighter. It works, but barely. This was when I thought there had to be a better way.

This was version 8 and fortunately most of the other versions died in CAD. The idea was simple enough just sandwich two pieces of aluminum together, drill a 1.75mm hole through at the junction, bung in a hot end heater somewhere and away you go. Then thermodynamics rears its ugly head and things start to get complicated. The heater is only 40W so the block had to be as small as possible. The heater is 6mm in diameter so 3/8” (9.5mm) base would be a minimum. A small 3/8 x 3/8 x 1” block would be ideal but would provide no support for the filament and it would just sag outside the heating block. So what I needed is a way to support the filament and keep it cool while the splice point reaches fusing temperature so a thermal break of some sort.

An air gap has to be about the best thermal break around but won’t provide any support for the filament so I had to fill it with something. Looking around the shop I fixed on the good old JB Weld. According to what I found on the internet regular JB Weld " can withstand a constant temperature of 500 °F (260 °C), and the maximum temperature threshold is approximately 600 °F (316 °C) for 10 minutes.”. This was great because I was only looking to hit between 200 and 230°C hopefully for under 2 minutes. Further reading on the internet I found that JB Weld was not recommended by the guys that over clock/hack their computers to mount heat sinks because it was only an OK thermal conductor. Excellent, good old JB Weld to the rescue ticking off all my boxes.

Water, petroleum, chemical resistant.

Resists shock, vibration, and extreme temperature fluctuations.

Can withstand a constant temperature of 500 °F (260 °C)

Can be drilled, formed, ground, tapped, machined, sanded, etc.

Not a great thermal conductor.

And best of all I had some.

Epoxied the two pieces that make up the bottom so that there was only about a 1/8” thick layer of JB Weld filling the gap and let it sit overnight.

I ran a couple of tests to see how high and fast the block would come up to temperature and it turned out that the placement of the heating cartridge was critical to performance. I got the best results with the middle on the cartridge sitting directly under the filament grove. Using an infrared thermometer the block gets up to 220 °C in about one and a half minutes. I then milled a 2mm deep groove to hold the thermistor. As you can see the JB Weld has discoloured from the repeatedly heating it but its still rock solid



It was time to see how things change with the extra mass of the cover on. So with the thermistor in its place and a k type thermocouple in the filament slot and another to read the temperature of the outer portion I was ready.


After repeated heating/cool down cycles things looked good. The addition of the top cover didn’t affect the times for the block getting up to temp. With the heater running up to 220 °C the outside never went over 30 °C so the filament is fully supported and kept cool except right on the heater block. My pyrometer also let me verify that my thermistor was reading within a couple of degrees of the thermocouple. Its time to add some hinges and see if this thing will actually work.


This was when things got a little freaky.

J-B Weld is water-resistant, petroleum/chemical-resistant (when hardened), and acid-resistant. It also resists shock, vibration, and extreme temperature fluctuations. J-B Weld can withstand a constant temperature of 500 °F (260 °C), and the maximum temperature threshold is approximately 600 °F (316 °C) for 10 minutes.

All good but I guess no one ever exposed it to molten PLA plastic. I wish I had a video of this one. As soon as the temperature reached around 190°C it spiked to over 300 on the OLED display. I thought something had gone south with my circuit but nope it was starting to smoke and the pyrometer was reading the same! No way that little 40w heater was capable of doing that. Pulled the plug and ran the cooling fan until things got down to a safe level. I had to pry the cover open and the JB Weld had gone soft and rubbery and partly fused to the heater.


The weather is still in the crap and were down to only about 6 inches of snow so on to version 9.

The next morning the JB Weld was back to being rock hard so I cleaned the remains off of the aluminum and started thinking of what I had on hand I could use as a DIY refractory. I could think of a lot of stuff I could use that would be perfect but like I said before “all the interesting stores are closed”. Digging around I found I had some of the furnace cement left I’d used to make my first casting furnace on the back of a shelf. It felt about half full ant I was shocked to find it wasn’t completely dried out. This stuff must be 5 years old!

Built it up the same way as the JB Weld and set it aside to air dry overnight. The instructions say to heat it to 200F, let it cool then up to 300F before putting it into service. Cycled it in the garage toaster oven (would NEVER use that thing for food) then ran it up to 500F to fully cure the cement. The problem with using that is that it gets most of its insolating properties from the air bubbles that form during curing. So even though it sands ok there is no way of getting the smooth finish that I want. I’ll want to completely redo this block when things settle down but as a temporary fix I just put a couple of strips of kapton tape on for now.

Bob, an interesting project, and intriguing results!

I have to confess that I use a very low-tech solution to this problem on my Bowden setup. As the end of one spool is pulled into the PTFE tube that leads into the extruder, I push the start of the next to butt up against it. Once the new segment has passed the extruder, I can go on my way; the new filament will push the old filament on through the PTFE tube that leads into the hot end. The downside is that retraction doesn't actually retract, since the new filament is not actually connected to the last bit that is going through the hot end ... but other than a bit of stringing resulting from the brief period that lacks retraction, it seems to work seamlessly.
Hi Andy

I have to admit that it’s a bit of a solution to a problem that isn’t much of a problem. The only real practical use I can think of for it is to use up the left over short lengths at the end of a spool. But it’s kept me entertained and given me something to do for a couple of weeks while everything is shut down. Now it just has to wait for some IC sockets to arrive and I can get it into an enclosure of some sort.
I have to admit that it’s a bit of a solution to a problem that isn’t much of a problem. The only real practical use I can think of for it is to use up the left over short lengths at the end of a spool. But it’s kept me entertained and given me something to do for a couple of weeks while everything is shut down. Now it just has to wait for some IC sockets to arrive and I can get it into an enclosure of some sort.

You say that as though a "practical use" were of importance. My wife keeps asking what practical use my model engine has ... I don't understand her question.

May I add my 2c worth? I was faced with a similar situation when wanting to weld the ends of a small urethane belt together from cord. My solution to this problem may be of help here.
My setup to hold the two ends of the round belt was to take an old hinge, and cut it so that only part of the curl that holds the hinge pin was used (a bit like a question mark). The hinge pin needs to be about the same diameter as the belt. I then screwed two of these pieces down onto a piece of plywood, so that they held the two belt ends in line with each other closely, about 3mm apart. The trick here was to clamp the one belt end down firmly, but to allow the other end to be just able to slide.
I then took my soldering iron and removed the copper tip, replacing it with a small (1.6mm flat) piece of sheet steel, shaped a bit like a comma. The tail of the comma went into the soldering iron, and the flat round bit was then held between the two ends of the belt.
The soldering iron was then warmed up and the"comma" put between the two belt ends, pushing gently on the free side. When the ends of the belt were hot enough and started to melt, they started to mushroom either side of the "comma". At this point, the soldering iron "comma" was withdrawn, and the free side of the belt pushed into the other fixed side and held until it cooled.
All that remained was to trim the flash off the mushroomed ends and I had a continious drive belt. The same could apply in this instance.

Nice solution :) Somewhat off topic, but would you explain in detail your heater and test setup?

Hi George thanks for stopping in.
HotEnd V8.2.jpg

I used a 12v 40w heater cartridge and a 3940 100K NTC thermistor. Both are basic hot end replacement parts and I had some left over from a 3 pack I’d ordered when I built my printer. The heater is switched with an IFR640 MOSFET and the fan is just on a 2N2222a (any NPN) transistor.
40w hotend heater.jpg

3940 100K NTC thermistor .jpg

This past winter my brother in law gave me a bag of ATMEGA328p microcontrollers so the control circuit is built around one of them with an Arduino UNO bootloader. I could have just used any Arduino and had it all up and running in a couple of hours but where’s the fun in that?

The sketch (software) is just the Thermistor Example #3 from the Adafruit Learning System guide on Thermistors Thermistor Thermistor by Limor Fried, Adafruit Industries. I added a couple of lines that says when the temperature reaches 205c turn the heater off and the fan on, a couple of buttons and an OLED display.

If you want any more info just ask.
Patience is a virtue but I’m having to be patient with so many things already with all the shut downs I decided that I didn’t want to wait for Canada Customs to decide if my IC sockets where safe to allow into the country. The last order I had coming from China was returned because the declaration stated that it was “Cell Phone Parts” and everyone knows that they put that on everything . . . . except these were Nokia 5110 displays . . . . from old Nokia 5110 cell phones . . . . but I digress.

I have some broken sockets so I just bodged together enough pins to make a 28 pin socket. It’s ugly but no one is ever going to see it, it’s not like I’m posting pictures of it on the internet.

Powered it up and no smoke so I loaded the program and happily everything was working so into the box.

I wanted it easy to use with consistent results. A reasonable cycle time. I have to say that was the hardest part but the total cycle time with a target of 205c is just over 3 minutes. I wanted it to have a small footprint and it’s 130mm x 110mm. All in all I’m pleased with the way it turned out.

Thanks Andy

Yep the case is printed. I lucked out when I built my printer that the CAD program I’ve been using for years exports STL files so zero learning curve on that end. I also found out that the only thing I hate more than flat square boxes for projects is cleaning out support material. The case is 5 separate prints tacked together with hot glue then set with 5 minute epoxy.
exploded view.jpg

I was printing some parts for my rototiller this morning (just a couple of knobs) out of PETG so I thought I’d try it out. It took me a couple of tries to find the right target temperature but it worked like a charm at 240c.


When I was using Slic3r for my slicing software I hated their supports, both the original Slic3r and the Prusa Slic3r. It always put a platform on top of the support columns, and that was a royal PITA to clean up. So I also avoided supports like the plague.

Then I went to Simplify3D and could turn off the support platform. Now I can put support columns just a couple of mm apart, close enough for the perimeter layers to easily bridge between the columns without sagging. The supports are a piece of cake to remove, and you can tell the software how many layers of gap you want between the supports and the first supported layer. If I tell it to leave a gap of zero layers between the supports and the first supported layer the supports need a little extra persuasion to remove, but it's still not too bad. One thing I have found is that I need to be sure and add extra supports anywhere the perimeter layer tries to change direction in mid air.

I had an older version of Prusa Slicer though, I think I had version 1.3.6. I just downloaded version 2.2.0 and it seems to do just about everything that Simplify3D does. It does gap filling on thin perimeters and it allows you to add supports. I haven't checked the set up for supports yet though to see what the options are available there, to see if you can get rid of the support platform. I just wish version 2.2.0 had come out a couple of years ago, before I shelled out the cash for Simplify3D.

Don, I've never been willing to pay for Simplify3D, but I have known it was the cat's meow for supports. I haven't tried recent versions of Prusa Slicer - need to do that!