Ultrasonic welding is the way to splice this stuff...


[A quick update:
The handle STL shouldn't be used because it presses against the rear end of the "sonotrode" assembly. This pressure has a similar impact to placing your finger on the cone of an audio speaker. The finger absorbs energy and muffles the sound vibration. Instead, hold or fasten the probe around its middle.

Learning as I go here... Ultrasonics is 1 complex field... wish I had a better math background... shoulda stayed in school. .]

I was not happy with the customary filament splicers. They are not easy to build and results are not great. But there was something in the back of my mind...

Ideally, to splice filament you wanna heat just a small section without melting anything past the splice, while the heating energy ought to be used quickly, and the heat source removed instantly. Additionally, it would be nice if the splice did not require any trimming or additional work. Anyone who has tried it knows what I'm talking about.

Then it occurred to me that industries have been using ultrasonics to weld plastics (and other stuff) for more than 50 years. It delivers the above benefits and more.

But where the hell does one find an ultrasonic welder? It turns out that some frequent household items might melt plastic if they're hacked correctly. Ultrasonic cleaners, ultrasonic fogger / mister / humidifiers. . These devices work at different cycles per second (hertz). Welding supposedly needs a specific range, commonly 15-70 Khz for plastics.

It so happens that most ultrasonic "cleansers" operate within this range. Pictured is a hand held, battery powered "Ultrasonic Stain Cleaner" I purchased at Harbor Freight several years ago. #96474.

Took it apart and hooked it up to a 0-18v 3Amp bench power supply. (It'll eat the 4 AA batteries within a minute since it draws over an amp. No wonder (almost) nobody sells them anymore.)

When everything's right the weld takes about half a second.
In an experiment, I held a piece of plastic against the piezo transducer (small round disk) of an ultrasonic Humidifier and it melted. They work at higher frequency but show much promise.

The large (~2" dia), bare transducer from a tank type parts cleaner has been removed and tested but it shattered into pieces before it melted anything. Ultrasonic transducers are somewhat delicate. Ceramic kinds are extremely brittle... They need strong support. Some must have a heat sink. The vibrations made this 1 self destruct. Watch the "exploded view". It was originally encased in epoxy and evidently had to be...

In particular, the thin metallic coatings needed to solder connection wires to the crystalline transducer peels off very easily. Then it's useless. (Electroless deposition of silver isn't difficult but is a lot of work. Better to preserve the original soldering surface).

The photos show a simple small Masonite jig to position and maintain the filament ends together. Vibrations cause everything to maneuver so spring clamps (not shown) are needed.

The end of the transducer "horn" includes a 1.75millimeter diameter half-moon notch filed into it. That notch prevents it from melting a level in the joint. Theoretically with the ideal horn and "anvil" shapes, an ideal round joint will be the result.

There is a jig STL to cut filament ends at 60 degrees (or is it 30 degrees? Whichever is the sharper angle...) to get an angled lap joint. Not the ideal joint, but it works well enough. A flat lap would be better based on what I have read, but it is not easy to do. .

Also included is the STL to get a handle to hold the transducer rather than leaving it mounted in the original plastic body. The circuit board (frequency generator) is laid bare in the picture. Box is in the background.


The welded joint is extremely strong. Actually, it's stronger than filament since filament polymer strings are oriented along its axis during extrusion, and it breaks easily cross-wise. With welding, the plastic is homogenous. There are no weak directions. See the examples of crossed filament breaks. The joint does not break.

There is a whole lot for this, and that I could write pages of info I have recently learned and found, but I'd rather people ask questions if something needs growth. Otherwise locate some suitable ultrasonic apparatus or appliance and hack along with me. I hope someone does because it took me 5 days to get this far and I'm running out of steam.

Related designs