Comment by DoingIsLearning | Hacker News Reader

Comment by DoingIsLearning | Hacker News Reader

Parent Full thread

DoingIsLearning

Superconductor though.

A semiconductor is a material that is somewhere in between conductor and insulator and varies depending on things like temperature or current direction. That is the material used in transistors and diodes.

I would say we are very far away even if this proves to be it.

First you would need to manufacture it reliably, then reliably without impurities, then reliably in some constrained 2d/3d geometry. Then you can start thinking about small footprint applications like IC design (chips and sensors). Perhaps then scale it to PCB design and RF applications like coplanar waveguides.

With that alone you would enter a new era in electronics with virtually no 'thermal noise' and no residual heat.

Beyond that (think large coils, motors, electromagnets) you would need a very large design step. As far as I understand this is still a very brittle ceramic, manufacturing very large or very long chains of this material would be unlikely. So the floating trains are probably a bit further away into the future.

I think the financial incentives will speed things up a great deal. Any business that can reliably manufacture it at scale stands to make billions. That usually spurs a lot of innovation.

Supposing this lead apatite if conducive, can we not use it for high voltage cables? Some absurd level of energy loss occurs in those.

I understand that there are current limitations inherent in superconductors, but is there no way to scale this?

The cost of losses in the cables may be high, but the cost of superconducting cables may be much higher.

Superconductors may be fragile and unsuitable for cables.

> Some absurd level of energy loss occurs in those.

It doesn't really - because we do the transmission at very high voltage, and the power loss is proportional to 1/V.

Power loss in transmission in the US is about 5%. In the transmission lines themselves it's only 2-4%. [1]

If you ran a power line all the way across the entire continental United States, you'd still get about 80% of the power out of the other end. The longest economically effective distance you can run an AC power line is about 2500mi, and DC around 4300mi. [2]

[1] https://chintglobal.com/blog/how-much-power-loss-in-transmis...

[2] https://en.wikipedia.org/wiki/Electric_power_transmission#ci...

"Depending on voltage level and construction details, HVDC transmission losses are quoted at 3.5% per 1,000 km (620 mi), about 50% less than AC (6.7%) lines at the same voltage."

https://en.wikipedia.org/wiki/High-voltage_direct_current

This seems pretty significant. We don't have much losses because we don't transmit energy over long distances. But now we could.

We definitely do, we just don't want to pay to string up wire thousands of miles, let alone superconducting wire. The difference between keeping 98% of the power that goes through a wire, or 100%, isn't the reason we don't do it. To quantify it further, the current US grid loses 5% to transmission losses which is just less than a cent per kWh.

Most power is generated in a centralized way anyways because it's much more efficient that way. The 'dregs' aren't connected because putting up the wire costs far more than the extra power yields. A few percentage points more efficient won't change the economics, especially if the wire is (a) lead and (b) dramatically more expensive.

3.5% per 1000km is respectfully, basically nothing. You'd get 85% of the power out of a line from SF to NY.

I'm not saying there aren't use cases for room temperature superconductors, I'm saying this is not one that's going to be top of the list.

> 3.5% per 1000km is respectfully, basically nothing. You'd get 85% of the power out of a line from SF to NY.

But why connect SF to NY - what's the advantage? What about connecting a place where it's midnight with a place where it's noon? That'd allow you to use solar arrays instead of local coal/gas/nuclear power plants.

I think you're missing the point of why we make power grids instead of simply having city local generators or stations.

The whole purpose of interconnecting power generation sources is to be able to accommodate for dynamic demand and ensure resiliency.

AC Power networks are sort of similar to how the internet works. The high voltage transmission lines are like the transit lines or "backbone" of the internet.

Those lines connect power stations which are sort of like ISPs in that they deliver the last mile power to the end user.

Our modern society basically instantly stops the second we are unable to meet demand for electricity, so we design these systems in a way where redundancy is supposed to be ensured.

This isn't always the case though. Texas is a great example of a completely messed up electrical grid that is insufficient to support its populous. It causes deaths in heatwaves and freezes almost every year now.

It's not just about transmission losses, though over time those add up. It's also about the cost of the rest of the infrastructure, which is in part because you're dealing with very high voltages and the various step-up/step-down requirements depending on what you're trying to do (long haul, local grid, last mile).

That's included in the 5% number btw. AC transformers are incredibly efficient.

In the loss calculations, not in the capital cost. Transport costs are a function of losses and amortized capital costs (and profits...). A huge part of power grid costs is because of all of the infra dealing with different voltages and transforming between them.

Totally. But I don't think we're anywhere close to knowing what the cost of LK-99 wire is to make a comparison.

That's true, we don't know what it will cost, we're not even sure it can be done at all... But HVDC as it is already in use uses very expensive cable as well so the gap could end up being smaller than you might think at first glance.

https://global-sei.com/technology/tr/bn84/pdf/84-10.pdf

This is state of the art, note the different deployment options, including pressurized ones.

Maybe they will make superconducting transformers instead of cables

One challenge is that these superconductors are ceramic. Brittle and inflexible.

Ceramics materials science has come a long way since the days of clay pots and flexible ceramics are definitely a possibility, depending on how thick you want them to be. Whether that is compatible with superconductivity is of course an open question but I wouldn't rule out a compound that is and superconducting and has a usable bending radius in at least one dimension.

The theory of this particular alleged superconductor is that it's a result of internal stress. I don't think that's compatible with bending.

Internal stress in one direction.

The material on the outside and inside of the bend radius is being stretched and compressed, respectively, along the direction of conduction (assuming any of the anisotropy stuff is more than speculation).

Yes, typically that results in a specific minimum bending radius to ensure material properties are not adversely affected.

This is already done, just not at great scale.

https://en.wikipedia.org/wiki/Superconducting_wire

For detailed information about one actually built and used: https://www.furukawa.co.jp/review/fr035/fr35_04.pdf

The enthusasiam is nice but there's a lot of NIH going on. I'd encourage people to research subject matter before thinking no one else has had similar ideas before.

It's "easy to make" in a sense, but the yields are insanely low (think 1/1000) or less of input materials. This indicates there are some variables that either are not controlled for or cannot be controlled.

That being said, its still early but it looks like LK-99 is not what we typically think about when we think of a super conductor. If we can figure out a good way to make it (with time we likely will), it will still have applications, just likely not high power transmission ones.

so a 100 GHz CPU could be not so far away

A 100Ghz CPU would be limited by other factors; mostly c.

That's true, but superconduction would change things from being planar to being cubic and that alone would give a huge boost to speed. Because one main limitation is to be able to get rid of the heat and building 'up' makes that very hard right now.

AFAIK CPU speeds are mainly limited by speed of light already today. There is tradeoff between time to fetch data from L1 cache (or register file), and their size. If you want to fetch the data faster, the cache has to be smaller (or the pipeline will stall), because the signal won't propagate fast enough to the cache. But smaller L1 cache also has negative performance impact, because more data has to be refetched from deeper caches.

Reply on news.ycombinator.com