The PETG adhesive test [video] | Hacker News Reader

The PETG adhesive test [video]

youtube.com

73 pointszdw35 comments

I'm very surprised to see the Polyurethane construction adhesive perform so well.

I noticed it blobbed over both ends. I wonder if the entire blob was helping redirect the load from tension into compression onto the ends. And/or it did a better job resisting the bending moment induced by his asymmetric setup.

I'm also not convinced that the other adhesives compromised the PETG. PETG, while not as resistant as Polypropylene, High density Polyethylenes, or Fluorinated plastics, it's still pretty darned resistant.

At the temperature he was testing at I would expect PETG to survive most acids, bases, and other solvents for days.

Maybe some of the glues were too rigid and robbed the PETG of its strength by preventing flexing?

MMA being at the top makes sense as it's a true chemical weld at that point.

Agree on all points - MMA being the winner is obvious.

One thing to keep in mind is that a lot of these PETG filaments are not true PETG's.

I didn't get a chance to watch with sound on, but curious if he verified it was really PETG

A lot of the PETG's are really PLA/PETG mixes.

Or event mostly PLA with some small PETG. Or something else entirely.

I actually have had filament claim to be 100% PETG, but it dissolved almost instantly in acetone, so not sure what it was, since even PLA doesn't really do that.

Often you can also tell it's not really PETG by the claimed glass transition temperature/etc, but ....

Sounds like an ABS or ASA blend ?

Yeah, that's a good guess. These days i don't have to figure it out any more because I mostly do nylon on printers meant for nylon :)

Interesting test and a topic I'm very interested in, but the whole product selection here appears to be very EU-centric.

In the states I've been using Weld-On 4 with some success for PETG-to-PETG bonds in high-stress applications: https://www.tapplastics.com/product/repair_products/plastic_...

Seems like a us-centric test is in order on our side of the pond

Great video, really appreciate the experiment design. I wonder if the same adhesive would perform well in situations that require them to flex?

I was helping a friend glue some PETG horns to a thermoplastic headband (long story) and some of the adhesives, particularly the superglue, worked well when the headband was stationery but cracked immediately when it needed to bend the slightest amount.

We ended up using Green Stuff 2-part epoxy clay as a physical "weld" between the PETG part and the thermoplastic (Worbla). AFAIK it held together but definitely not an ideal solution :)

This is really cool! Don't all the ladder test cases that achieve >~50 simply snap the PETG implements though? So isn't the test capturing the force to snap PETG at that thickness vs. the actual bonds?

Yes. There are three types of bond failure:

- adhesive failure: the adhesive splits from the substrate

- cohesive failure: the split occurs within the adhesive

- substrate failure: the split occurs within the substrate.

Whichever of these that is weakest limits the maximum strength of the bonded item.

Yes, but there is a reproducible difference between when different adhesives break the PETG.

He concludes that those adhesives are causing the PETG to become brittle, and you should prefer others which don’t.

I don't have a ton of experience gluing PETG--the one I've used is "PETG Gloop", which I don't think he tested. I wonder if there's a close analog, or where it would fall in his rankings.

(PETG Gloop SDS is available here, if you want to see the chemical components: https://www.3dgloop.com/sds )

Missed PVA. That works fine for PLA and is dirt cheap.

I think Ponal Holzleim Super 3 he tested is PVA based wood glue. MSDS states it based on polyvinyl acetate dispersion.

Interesting Auto-Dub feature.

Yes, that's cool, however, I would like not to be forced auto-translations from languages I understand.

I am French and this is my primary language, but I understand English and Google knows that (it is in my account preferences). So why do I get videos auto-translated from English into French, with a significant loss in quality. It is particularly bad for titles, the translations are terrible and I can't get YouTube to show me the untranslated title. I probably could set YouTube to use English instead of French, but then, I would have French videos translated into English, which is even worse.

It's absolutely infuriating. It could be such an awesome accessibility feature if it wasn't force-enabled by default. But what we get instead is YouTube engineers assuming that everyone just speaks a single language and needs auto-translations on every video. For every channel where this feature is enabled, I have to keep switching back to the original English audio track on every single video because YouTube knows that just because my primary browser language is set to German (the secondary language being English, btw), I am incapable of understanding the original audio track, of course.

I believe you can disable it clicking on the gear.

On every. single. video. again. and. again. They don't remember your choice.

And it keeps the auto-translated video title, so if that one is garbage there is extra work involved in figuring out what the original title meant to say, too.

Yep, it's a terrible materialization of a theoretically awesome idea. Both because of these details and because of the robotic tone of the voice compared to the demos they had released.

Click the gear icon.

There are so many realistic text to voice project shared on HN. Why YouTube is using robotic voice, and it sounds irritating and I can't listen to it more than a minute.

It's been years since Google demoed their shocking AI voice calls, and still this seems to be the state of things. It's hard to separate the words from the robotic voice here but the state of Google Translate has also been really sad the last few times I've tried it, given the fantastic translation that most LLMs can provide these days.

The only thing that makes sense is that good models require a lot of compute, probably more than the transcoding that happens when you upload.

It's almost certainly because of the scale at which this type of thing needs to run. There are a lot of YouTube videos.

I don't know, I would prefer such things to be slightly robotic over them being in the uncanny valley.

This guy's German is very clear, and it's a joy to listen to the presentation. What is the accent? I can see he's based near Münster. I can recommend listening in original language if you know it even a little bit.

It's a bavarian/Munich accent.

Yes, but I would prefer the original audio with translated captions. Why isn't that possible? If they can auto-dub, it should be even easier to give us translated captions.

Same, disabling it and looking for subtitles was the first thing I tried to do.

It is, click the gear icon.

No, only German subtitles. No auto-translated to English (or some other language I would understand).

Maybe auto-translation isn't available in your country. It's there when I watched the video.

Seriously. We get so jaded by problems and over-hype of what we collectively call 'AI' but this is something that would have seemed like impossible futuretech 10 years ago.

Indeed. YouTube is rolling this out to more and more folks, and it gets really interesting as you can kinda produce videos in your language but still have an international audience. The quality is not outstanding as the voice is a bit robotic, but still... In my channel even if I don't yet have the feature enabled, just because of subtitles auto-translation, I'm seeing more and more foreign subscribers.

It's interesting but I wish I could turn it off. would be nice to have a button "dub this for me" instead of being forced to have all the content translated into 1 language, which is infuriating if you understand multiple.

Edit: aha you can switch back to the original audio track

Reply on news.ycombinator.com