Hacker News Reader

Comment by nonrandomstring | Hacker News Reader

I think there's a difference between them being "attracted" to the short range field, and being disoriented as they climb over the cable. Maybe it's the heat or inductive heating effect they like. EM shielding the cables would make them too heavy. Switching to DC transmission might fix it, but the cable loss would be higher and the cost of rectifying and remodulating to AC would be high.

Two thoughts come to mind.

More cables in parallel carrying less current, which gives the added advantage of redundancy.

Short breaks in the flow, like a railway crossing for crabs, for 5 minutes or however it takes them to get unstuck and go along their crabby business.

View on HN

>Switching to DC transmission might fix it, but the cable loss would be higher

AC loses due to reactive impedance and skin effect should always be higher than DC transmission. DC might make sense if the generation is solar based, as it naturally generates DC, although converting it to higher voltages must have an AC step.

The highest capacity sub sea transmission cables are DC now I think. There is an interconnect between Ireland and France that is DC if memory serves.

I think it's only efficient for the largest capacity lines tho

AC cables are just more expensive per km, I think due to skin effects.

That and the enormous capacitance between a long cable and the sea.

DC is also used because you avoid grid synchronization issues, and this is generally the trend for long distance connects now in general. It means the grids can have frequency divergence without affecting the transmission line itself.

The issue with DC transmission in subsea cables is that the XLPE insulation can become polarized and - from what I remember - have water drawn into it due to the consistency of the field.

You could probably flip the current at periodic intervals to deal with this.

You're talking about flipping the polarity, going occasionally from conductor-positive to conductor-negative (the shield must remain at ground potential.)

If I remember correctly that actually makes things much worse if you've operated at one potential for a considerable amount of time.

You could operate the cable at AC ~ 0.1Hz but your ability to actually transmit MWH would suck.

And DC avoids problems caused by induced currents due to geomagnetic storms (you get a voltage shift, but that's no big deal, compared to what happens to transformers.)

Seems unlikely to be a problem undersea, no?

The induced currents flow in the ground (and water) which cause big voltage differences at either end.

Whatever, the ground currents penetrate very deep into the earths crust thus equally effecting buried cables.

grid synchronization 'issues' are a feature, not a bug - by very, very slightly varying how hard to you try to 'push' out of synch you can manipulate power flows at will. This is dramatically more difficult with DC. That, and ease of voltage level manipulation were the reasons so much transmission was AC in the first place!

No they're issues: if it was easy, then no one would bother with HVDC - HVDC exists because it reduces losses and vastly eases grid coordination because conversion to and from AC at either end is handled under local control and can be totally independent of the source/destination.

Its the difference between just shifting the frequency or phase on your local control clock at the substation, and trying to coordinate with another state/territory/nation's entire AC generation infrastructure to keep everything in phase.

Informative but entertaining video on high power DC transmission from Electroboom. To my recollection it’s generally always better transmission wise except the DC-AC and AC-DC converters are both expensive and physically large.

https://youtu.be/DFQG9kuXSxg

Both size and cost are dropping pretty quick with the development of better MOSFETs - the same ones that are soon to get very cheap because they're used in electric car motor inverters. I predict that in 100 years the majority of big power links will be DC (although we'll probably still have an AC grid too)

> More cables in parallel carrying less current, which gives the added advantage of redundancy.

Every time we have a problem with tech the answer always is "just add more", idk how to feel about that

Or more abstractions. Where the original problems that we were trying to get away from, resurface in new ways, and previously solved problems have to be reimplemented.

Adding more of something is often the simplest and cheapest way to fix problems. Example: Adding a few more VMs or containers costs almost nothing. Paying developers to eke out 10% more performance is expensive.

Another example: Microcontrollers that have loads of pins (e.g. 100) and support many different peripherals (i.e. protocols like I2C) are expensive (say, $10-20). Little ones with a medium amount of pins (e.g. 56) and support for limitless protocols are cheap ($1). So if your PCB needs just a few more GPIO pins to accomplish something it makes more economic sense to just "add another" little MCU rather than "go big". Even if it makes programming the thing slightly more complicated.

Combine the two? Several smaller cables with a cycling subset powered at any given moment.

You are claiming that DC losses are high, can you cite any source.

Probably only historical ones. In the past it wasn't easy to get high voltages with DC, so losses where higher. These days we are better at changing voltages, and so DC losses the same as AC with similar voltages (skin effects still can be a small difference).

I understand we can actually run DC at higher voltages than AC, though I'm not enough of an EE to explain why. IF this is true, then DC losses would be less than AC.

DC-to-AC losses used to be high (historically) because of inefficient inverter technology. These days we have 90-95% efficiency inverters (that are built to match their HVDC voltages).

It's interesting because the reasons why we built everything using AC in the past don't really apply anymore (well, most of the time). DC is more efficient for most domestic and industrial electricity utilization. Things like lighting, air conditioning/heating, refrigerators, etc are all more efficient if fed from DC. Think about it: How many "wall warts" have we got in our homes these days? All our LED light bulbs have built-in inverters and that decreases their efficiency. Our PCs use DC for literally everything internally. The list goes on and on.

I had also read a while back that grid-scale DC power transmission could be more resilient because the power transformers aren't as expensive. It's like $8 million for one of those giant high voltage AC transformers they have at power stations whereas with high voltage DC the equipment is cheaper but you have more of it (or something like that; I wish I could remember the details but this was an explanation given to me from a DC fanboy lineman over 10 years ago). The idea being: If bad weather breaks your giant $8 million AC transformer that's a real expensive pain in the ass but if it breaks a handful of small DC transformer components that's a quick (and cheap) fix.

> LED light bulbs have built-in inverters

Think you meant rectifiers here.

Can you clarify the difference?

Generally Rectifier = AC to DC and Inverter = DC to AC

Sorry, but no. An inverter uses switch-mode technology to enable voltage conversion. Usually this allows using small high-frequency transformers, rather than large 50/60Hz iron cored units. Which is why so many wall-warts use Inverters to step the voltage down.

Yes, switching power supplies can be used to go from AC to DC, but those are not inverters. They're a form of active rectifier.

Perhaps you learned it differently, but according to any source I can find, they're defined as I said.

Inverter: DC -> AC

Rectifier: AC -> DC

Edit: whoops I'm late to the party

Actually not really... A single-phase-AC-sourced rectifier does need a considerable output capacitor, though, and it will have a poor power factor with severe harmonics unless it spends extra on PFC which would not be necessary with a local DC grid supply (say, 320V DC to replace 230V AC systems to match peak voltage; yes the higher RMS voltage results in (anti-)proportionally reduced current, allowing cheaper/thinner wires or less loss).

Many LED bulbs are a simple diode rectifier circuit, essentially with a bunch of LEDs stacked to add up to a total 120V bias voltage. These will blink at 60Hz or 120Hz depending of if they implement a half-wave or full-wave.

More complicated bulbs use an active rectifier / switching power supply instead.

Those are kinda shit, though. Also, the ones you're thinking of should be capacitive dropper circuits, if I'm not mistaken.

There's also the electrolytic-smoothed bridge rectifier variant. BigClive has a lot of teardowns of such light bulbs (and non-bulb-shaped alternatives) with circuit diagrams and explanations. They're really good for learning about the various "cheap LED bulb" designs.

AC is easier to switch than DC because you get less arcing. Good demonstration:

https://www.youtube.com/watch?v=mQpzwR7wLeo

Not if you use solid-state switches.

Actually, they're much better than "just" 95%. 98-99.5% is realistic with at least non-isolated converters for each end.

And yeah, a DC grid needs active converters at connections points anyways, and the small voltage changes needed for steering power flow across redundant paths is also relatively cheap.

A big issue is just that a constant-power SMPS with an input voltage range/tolerance inherently offers a negative differential (aka incremental) resistance: within the supported input voltage range, reducing the input voltage causes the current consumption to increase inverse-proportionally (typically even a tiny bit worse because internal losses in the SMPS increase with current, requiring even more current to make up for the slight drop in efficiency (think 98% to 97%, for a sense of scale)).

In AC grids phase is comparatively easy to stabilize, and constant-power loads already need PFC that (for single-phase rectifiers) has to emulate a normal resistor who's resistance slowly drifts to accommodate changes in output power consumption. The only real issue is that this method requires an intermediate stage with large capacitors to feed the output stage while the input stage stops feeding near the zero crossing (the capacitors get smaller if you do away with the zero crossing (DC grid using the emulated resistance for stabilizing the grid), but the stages remain, along with their ~1% inefficiency per stage).

The main issue with LEDs is that you need electrolytic capacitors to buffer the zero crossing of single-phase AC, and electrolytic capacitors dry out/corrode when operated at high temperatures. So unless you want the LED PSU to have a fan, those capacitors will severely affect how tightly integrated you can build a non-low-power LED light bulb.

And yeah, especially at a local scale DC grids are very useful, because you can skip expensive inverters and conversion losses for both battery(-backup) sources and solar panels/fuel cells. You would typically still want an MPPT for a string of solar cells, but that's a compact and rather inexpensive DC/DC converter. Batteries only need a way to accurately limit the charge voltage/current per-pack, to prevent the voltage drop along the grid cables from affecting the maximum charge voltage/state of the pack's cells. Also a deep-discharge protection, but beyond those, a (mostly) solar-fed local DC grid can just let the MPPTs feed as much as they can up to a target voltage, from where they start to gradually throttle output current to only produce as much as is being used (once the batteries are full). If the batteries start to reach sufficiently low charge state, load shedding or alternative power has to be used (say, via telling PCs to reduce/disable Turbo Boost and similar, or lights/screens to limit brightness).

More cables could disrupt the sea bed more during construction and cause more harm to plants and animals.

Maybe a barrier to prevent underwater critters from getting too close to the cable. Some kind of tubular fence?

crabby business <3

Reply on news.ycombinator.com