Solid state circuit breaker by Atom Power

[HackedFridgeMagnet]

Anyone got any more info on these?

Apparently.
Just got UL approval.

Can break 100kA.

https://www.popularmechanics.com/technology/infrastructure/a27557804/digital-circuit-breaker/

[SG-1]

Information like this ?
https://pdfs.semanticscholar.org/d79c/6d6c1a8c8fa1e620edae3865076f773b6803.pdf

[SG-1]

More here:
https://spectrum.ieee.org/energywise/energy/the-smarter-grid/atom-power-is-launching-the-era-of-digital-circuit-breakers

[NiHaoMike]

A lower cost version for bench use would be nice, especially for those who regularly work on power supply circuits.

[T3sl4co1l]

I've done that before:
http://www.seventransistorlabs.com/Fuse/index.html

From what scant little info there is in the article, it sounds like it's comparable, i.e., a fast-acting electronic switch.  Unfortunately this isn't very useful for mains applications where surge loads are a necessary requirement.

They may be handling mains voltage surge.  SiC is available into the kV range, though you still have the Vds-Rds(on) tradeoff that all MOSFETs have*, so they probably need quite a lot of die area to keep resistance and dissipation reasonable.

*I don't remember if they're doing SuperJunction designs in SiC, but that would be important in a switch like this.  The functional difference is, older generation VDMOS scaled as Vds ~ Rds(on)^2, i.e., really terribly (double Vds --> quadruple Rds(on) for the same die area), whereas SJ is proportional, so the power switching capacity scales with die area, more-or-less independent of Vds.

They're definitely not handling fault current as a linear regulator.  The energy handling of semiconductors is just pitiful.  More likely they'd dump the excess into a MOV stack.


My electronic fuse, does just what it's asked to do: actively limit current in the domain of ~us, and turn off (in ~us to ~ms) when the current goes above the setpoint.  Unfortunately it's rather annoying because most DC loads have more capacitance than this thing can charge, even if you rapid-fire restart (keep tapping the 'on' button).  So it nuisance trips, and it's really hard to use.

So, I fear that a semiconductor "breaker", of useful size, will suffer a similar fate.  At least with conventional loads.  Maybe with all smart loads (with inrush protection or precharge circuits, and PFC, and..), it would be feasible, but that's asking a lot even for a new built home in a trendy place.


So, later I made this,



which has a much more respectable capacity: 30V 20A nominal rating (maximum 40V drop), active current limit for 150ms duration, and thermal protection.  That's enough charge (a whopping 3 coulombs) to start pretty much anything (nominally, 100mF at 30V?).  It's also bidirectional so it can be used on bipolar DC sources, batteries (no worry about limiting charge or discharge currents), that sort of thing; but, probably not AC, due to the capacitance.

The main downside is the amount of noise thrown off (the filtering is not exactly FCC Part 15 grade), but it's only generated during fault conditions so it's not a big deal.  I mean, that and the cost.

Since it does have capacitors and inductors inside, it's not as ideal of a switch as the other one, but you can't have everything.

Mechanical switches are damned hard to beat, offering huge on/off ratios even at high frequencies, and handling lots of joules of make/break energy (depending on design, of course -- mains breakers are designed to start an arc and launch it into a heatsink, until zero crossing; doing that at 10kA and hundreds of volts, for about 10ms, is a lot of energy!).


Going back to the breaker, I would think it much more likely that it's just a pair of solenoids, a bit of thinky-stuff, and an utterly conventional (mechanical) breaker.  That's entirely possible, and affordable, with off-the-shelf parts, and doesn't expose any semiconductors* to mains surge or fault conditions.  It doesn't sound nearly as exciting, which may be as much of a reason not to discuss its internal workings.

*Well, still the solenoid drivers, but that's a lot easier to manage.

Tim

[james_s]

Call me old fashioned but I'll stick to my plain old thermal/magnetic circuit breakers. I don't need a breaker to be 3,000 times faster, in fact I'd think that would cause a lot more trouble in the form of nuisance trips than it's worth. A breaker just has to be fast enough to prevent the wiring from catching fire, while being able to interrupt the maximum available fault current. It also needs to be able to be switched off and tagged with 100% reliability that it won't turn itself back on under automated control or by failure.

[Berni]

So what happens with these when a big thunderstorm comes by?

Lightning hits something near by and induces a few kV on top of the mains for a few microseconds. Does this blow up the control circuit and leave it inoperable? How about it survived the first hit and turned off as designed, but then a second hit comes around and hits it in the disconnected state where all those kilovolts are suddenly applied across the semiconductor switch. Does the switch break down and let the surge pass on trough the house? Does the switch perhaps now fail into a shorted state? In that case once the electrical company restores power to the street the breaker is permanently on, maybe that surge also blew up whatever is connected to that mains cirucit and it failed short too, now there is 100s of amps flowing with nothing to stop it, making the wires going there glow red hot and start a fire.

All of this just so you wouldn't have to walk to the electrical cabinet every 3 years or so when a occasional fault trips a breaker, in witch case you probably also have some problem in the house that caused it to trip and you have to investigate manually anyway.

Do they have a statistic on the number of deaths caused by too slowly operating circuit breakers? The last time i checked a 10A circuit breaker will just as happily let 10A flow into a human being or a toaster. These things save houses from burning down, not people from behind electrocuted.

[Jeroen3]

Atom Power’s digital circuit breakers are 3000 times faster and 100 percent safer than mechanical circuit breakers.

So it's only faster?

And I don't think internet of things circuit breakers are a good idea.

The http requests are not
authenticated unless the credentials are correct, hence restricting any kind of
communication from a malicious source.

What about https?

[T3sl4co1l]

So what happens with these when a big thunderstorm comes by?

Lightning hits something near by and induces a few kV on top of the mains for a few microseconds. Does this blow up the control circuit and leave it inoperable? How about it survived the first hit and turned off as designed, but then a second hit comes around and hits it in the disconnected state where all those kilovolts are suddenly applied across the semiconductor switch. Does the switch break down and let the surge pass on trough the house?

Like I said, it's entirely possible to be rated for the full 2.5kV surge you normally expect on mains.  It would take a lot of MOSFET to realize, though.  An inlet surge arrestor would be obligatory.

More likely, if it is still a semiconductor switch, it's got MOVs on both sides, limiting peak voltage to about 900V, so that 1200V FETs can be used (1700 or 2000V if simultaneous and opposing surge is also to be protected against; that might be relevant under EMP conditions -- very close lightning strikes, say).

Tim

[james_s]

Even with good protection the idea makes me nervous. In my experience, semiconductor devices almost always fail shorted, and they do fail occasionally even in the best most conservatively designed equipment. Electromechanical breakers on the other hand can fail shorted but it's rare enough that I've never personally encountered it, and once one is shut off it's virtually impossible for it to fail in such a way as to turn itself back on.

[schmitt trigger]

Same experience here....Semiconductors fail short.

[HackedFridgeMagnet]

I think the IoT aspect is not really interesting as you could do that with existing technology.

But if this breaker technology isn't bullshit and can be done at a reasonable price point then you would have to assume that eventually the price point would come done and the reliability would come up to the point where it is the superior breaker, and also superior contactor.
But I've got the feeling I will be waiting a while.

In the meantime I am still wondering what is in the box.
Tims conjecture seems plausible. Though I thought it said somewhere 100kA breaking capacity. That would be impressive for a semiconductor.

More likely, if it is still a semiconductor switch, it's got MOVs on both sides, limiting peak voltage to about 900V, so that 1200V FETs can be used (1700 or 2000V if simultaneous and opposing surge is also to be protected against; that might be relevant under EMP conditions -- very close lightning strikes, say).

[T3sl4co1l]

Yeah, 100k not gonna happen, more likely it active-limits at some moderate value (or the line inductance serves as adequate limiting), and the claim is simply that it was tested on a generator capable of 100k S/C.

Actually, 10us and a typical line inductance of say 100uH would be very reasonable (that's a mere ΔI = 40A), I doubt it does any active limiting at all.

I measured a similar waveform here, on the electronic fuse device, by switching in a 30V supply (of very low source resistance) while the fuse is active (on):



In the first microsecond, current rises to 120A (this is -20A/div equivalent scale).  Active limiting kicks in and limits it to about 60A flat band, before shutting off.  A small undershoot occurs as the stray inductance discharges into the TVS diode.  This was with a mere 150mm twisted pair lead, or about as many nH.  So you can see, inductance is very important on these time scales.

(The corresponding waveform for the switching limiter circuit is much less remarkable: there's a slight overshoot on the leading edge (due to response time), then there's a somewhat lumpy (switching ripple) flat square pulse about 150ms long, then it stops. )

Tim

[jbb]

Maybe it’s a hybrid with a semiconductor stage to limit current and a mechanical disconnector to make it stay disconnected.  I expect UL were quite thorough testing this one.

Let’s hope Big Clive gets his mitts on some of them .

They could be very good news for DC installations. DC is hard to break but otherwise attractive for a lot of modern equipment (eg data centres) that have switch mode supplies in anyway.

Finally, I wouldn’t trust a purely semiconductor breaker to keep the circuit disconnected while I work on it...

[Jeroen3]

Even with good protection the idea makes me nervous. In my experience, semiconductor devices almost always fail shorted, and they do fail occasionally even in the best most conservatively designed equipment. Electromechanical breakers on the other hand can fail shorted but it's rare enough that I've never personally encountered it, and once one is shut off it's virtually impossible for it to fail in such a way as to turn itself back on.

Somewhat larger circuit breakers are common to fail closed. The mechanism get jammed or it closed with a large inrush welding the contacts.
So yes, I've heard it's quite the experience arriving at a grid support genset where the breaker failed closed and the engine shut down. (hint: generator keeps running)
In your home or office installation the mechanical circuit breaker will never fail close since they are underrated. Sometimes an GFCI gets lazy and won't open, but that's due to the trip coil being jammed.

I can see some advantages of this system. You can configure them slightly below the mechanical breaker profile and remote monitor and intervene if needed. Since mechanical reclosers are big and expensive.