Hollow Core Fiber (HCF) | Hacker News Reader

Hollow Core Fiber (HCF)

holightoptic.com

I was going to do the "well akshually..." about HCFs because this article is so light on details. I used to work in a related field many years ago, and the attenuation of these fibers relegated them to niche applications (short-lengths that needed the odd properties of a hollow core, like ultra-high power transmission, low dispersion, low nonlinearity, etc.) They were orders of magnitude too lossy for long-haul telecommunications applications at the time.

But a quick search proved my knowledge was not up to date - here is the paper linked from the Laser Focus World blurb someone linked above. Looks like HCFs have achieved loss parity with traditional single-mode fused silica fibers, at least in the lab. Pretty amazing and cool if you ask me!

https://eprints.soton.ac.uk/470034/1/OFC2022_PD_Jasion_submi...

Do the wave modes offer higher velocity?

Yes, basically most of the electromagnetic wave is in the hollow (gas-filled) core, which is going to have an index closer to 1 (as opposed to ~1.4 for silica fibers.)

Oh, wow! Thanks for clarifying. I get it now! (Former optical engineer, but not in fiber optics.)

A key detail I found lacking in this article was the manufacturing cost difference. It mentions that it is more expensive, but no indication of how much more expensive.

A lot more expensive right now. But silica fiber manufacturing has been optimized like crazy for 50 years, so that's not a surprise.

From a high level, the preforms are a lot more complicated, and I imagine the draw process might be more unforgiving (in terms of size fluctuations of the cross-section), but I don't see anything that is going to be prohibitive or that will make the fiber itself 10x more expensive in the long run. Most of the cost of an undersea cable isn't in the silica anyway...

> ultra-high power transmission

Does this mean that such fibers can be used with high-power lasers, like those used in tools and weapons?

Yep, that is what funded a lot of this research early on.

Hollow-core optical fibers may have a bright future [1], along with another HN thread from 8 years ago [2], I have longed for these cables where we could cut Japan to New York latency by 50ms.

Unfortunately we dont seems to be anywhere close to it.

[1] https://www.laserfocusworld.com/fiber-optics/article/1417001...

[2] https://news.ycombinator.com/item?id=16194973

No... hollow core fibers are being installed as a test in some data centers now/this year. Who knows how many years it will take to (re)make an oceanic backbone (and the prices…)

I'd also argue that the 50ms reduction time benefit with HCF has waned with the introduction of HFT (High Frequency Trading) over LEO links with lasers.

This is fascinating. As an engineer, my mind always goes to: "what are the practical challenges":

* How do you splice without crushing the core?

* How do you terminate/splice and maintain vacuum?

* How do you terminate at all

tldr: lasers (angry light)

How do you make the fiber in the first place?

> * How do you splice without crushing the core?

Pressurize during the slice, and slice quickly before the pressure drops. Not hard over that very short distance.

* How do you terminate/splice and maintain vacuum?

Gas-filled, not vacuum.

* How do you terminate at all

See above.

Not a comment on the article, but the linked URL makes me vaguely uncomfortable.

That fullwidth question mark really makes the URL stand out, not in a good way.

I think that it was good that it stood out to be honest.

Slightly too clever CMS. Like Dolphin replacing slashes with the Unicode fraction slash.

I hate it. So... much.

Curious about how difficult would be to splice it.

Good point. I have no idea, but probably the tools will need an upgrade!

yea hand splicing in the field is already super rare today. I imagine this cannot be hand spliced at all.

Considering the number of fiber splice trucks rolling around I think your supposition is incorrect. Fiber splicing happens thousands of times every day. Fibers break and get cut all the time. Tt's not common in data centers, but once you leave those walls fiber splicing happens everywhere every day.

Also just upgrading the connecting bits. Like the muffle/joint in the ground, for more 'speed',

to enable more branching out, or both. Seen yesterday night, had a good chat with the two guys.

I always stop when I see them, to get a better picture of what is where, who owns it, and so on.

Sometimes they even let YOU do the splicing. Which is actually easy, because it's fully automated.

Imagine something like a small suitcase, or a shoebox with LCD-screen on top. You insert both ends into slits on both sides,

the 'box' grabs and aligns them, cuts and repolishes the ends, melds them together, and ejects the fibre through a slot in the front.

Then you sleeve that with tape, insert it into a muffle, apply some hardening foam, sometimes gel, like hot-glue, and close that muffle like a clam shell.

Done.

Yeah, splicing is a lot easier these days.

I've personally spliced about a dozen optical fibers and I don't even work in telecom :)

Not exactly by hand though - all the prep is done by hand but the actual alignment and fusion steps were automated by the fusion splicer machine (about the size of a small toolbox)

https://techcommunity.microsoft.com/blog/azurenetworkingblog...

Suggests it can be to some extent.

So what's the application for data?

Would this provide for tighter channels in WDM networks (with less spacing between channels) since dispersion is less of an issue? Might be out of my league on this but I thought DCMs were a work around for dispersion at distance.

Maybe faster packets for teleprotection control systems?

Either way, cool to see!

How do you evacuate these fibers to any decent vacuum?

Sticking a vacuum pump on one end and waiting presumably won't work because gas will take a really really long time to diffuse along a multi kilometer long micrometer wide tube...

My uneducated guess, fill the cavity with a reactive gas, then apply heat in one part of the process to react the gas with the walls, leaving a vacuum. Part of the article mentions silver cladding on the inside wall, which would be a getter for some gasses, with the downside that it is slightly tarnished.

Or, the entire process is performed under vacuum.

More likely, that part of the manufacturing process isn't fully solved, so they just have pumped samples.

You produce them in orbit. Like ZBLAN. Actually hollow cored ZBLANS seems like a nice idea, to a layman like me :)

I wonder if we'll see more power-over-fiber products with this kind of thing?

The article is full of AI slop, sorry.

Reply on news.ycombinator.com