AFCI's - How do they work??

[K8MHZ]

Rather than skew off a GFCI thread, I though I would post some info about how the cursed AFCI's actually work.

I even found a video from Eaton where they used real sparks to demonstrate the breakers. You may want to fast forward to about the 5 minute mark if you want to skip the basic electricity lesson.

http://www.youtube.com/watch?v=ihforRdL-TU

The crux of the biscuit is that an arc created by a fault will have a current symmetry that differs from an arc created by an intentional load.

http://www.pfeiffereng.com/The Arc Fault Circuit Interrupter.pdf

The above link also has a block diagram of the workings of an AFCI.

Here is a link to Square D's documents about AFCI's

http://products.schneider-electric....bination-arcfault-circuit-interrupters/#tab=2

Here are some basics.

Most household breakers have a 200 amp short circuit rating, meaning for a very short period of time, they can carry 200 amps and stay closed. This is actually needed to assure that devices with large amounts of in rush currents don't trip the breaker.

The average amount of short circuit current available at the average receptacle is 75 A. This means a direct short at a receptacle will take part of the time curve to trip the breaker.

One feature of an AFCI is that they target a current below 75 A for an instantaneous trip.

The 'signature' of an arc is what it's pattern would look like on a scope. It looks like arcs coming from faults are very asynchronous and have characteristics that arcs from say, the brushes of a motor have. The signature of an intentional load, no matter if a bit messy or not, will have a repeating pattern from start to finish. Fault arcs will have more of a random pattern and usually will exhibit decay before they burn themselves out.

Now, the 75 amp and below limit is for a hot to neutral fault. The amount of current that in a series (broken hot or neutral) is already limited by the load, so the amount can be below 15 or 20 amps, for instance, and still make a dangerous spark. A combo AFCI is one that will shut down on both a series (current limited by load) arc and a parallel one (hot to neutral). It doesn't mean it has a GFCI built in.

 

[iwire]

All assuming they are being entirely honest with us.

 

[don_resqcapt19]

All assuming they are being entirely honest with us.

Why would we even think that they are doing that? These are the people who said, in the original AFCI proposals, that they had a device that does what they now tell us the combination type AFCIs will do. Those original proposals were some 13 years prior to the combination type AFCI hitting the market.

 

[K8MHZ]

All assuming they are being entirely honest with us.

That's why I liked the 'arc under glass' demo.

Instead of just showing graphs and waveforms, actual arcs were used.

Years ago, when I installed my first AFCI, I put it to a test. Using a heater as a current limiter I created arcs by touching wires together for a series arc test. No matter what I did, the AFCI would not detect the series arc. I did not test from neutral to hot, since I didn't wire the entire house, just an added bedroom.

 

[ELA]

Good to know someone has finally figured out how these things "actually work" :

http://forums.mikeholt.com/showthread.php?t=148992&page=3&p=1439490#post1439490

 

[Fulthrotl]

That's why I liked the 'arc under glass' demo.

Instead of just showing graphs and waveforms, actual arcs were used.

Years ago, when I installed my first AFCI, I put it to a test. Using a heater as a current limiter I created arcs by touching wires together for a series arc test. No matter what I did, the AFCI would not detect the series arc. I did not test from neutral to hot, since I didn't wire the entire house, just an added bedroom.

i did the same thing with a chief inspector for a city near here....
you can pull an arc almost big enough to weld with, and not
trip that puppy.... hook up a load and touch the wires, making arcs....

the only thing that changed was the $50 that disappeared out of my
pocket when i bought the AFCI breaker... :rant:

 

[don_resqcapt19]

....
Years ago, when I installed my first AFCI, I put it to a test. Using a heater as a current limiter I created arcs by touching wires together for a series arc test. No matter what I did, the AFCI would not detect the series arc. I did not test from neutral to hot, since I didn't wire the entire house, just an added bedroom.

The AFCI people have never said and still don't say that an AFCI will detect and clear a "series arc". There is not even anything in the UL AFCI standard that talks about detecting and clearing a series arc.
The only way an AFCI can detect and clear this arc is when the series arc progresses to a parallel arcing fault or a ground fault (if the AFCI has a GFP circuit).

 

[ptonsparky]

So the arc that my wife's vacuum sweeper creates every time she attempts to use it on afci protected circuits is parallel and not series. Hmmm.

 

[don_resqcapt19]

So the arc that my wife's vacuum sweeper creates every time she attempts to use it on afci protected circuits is parallel and not series. Hmmm.

At least that is what the AFCI thinks it is

 

[K8MHZ]

The AFCI people have never said and still don't say that an AFCI will detect and clear a "series arc". There is not even anything in the UL AFCI standard that talks about detecting and clearing a series arc.
The only way an AFCI can detect and clear this arc is when the series arc progresses to a parallel arcing fault or a ground fault (if the AFCI has a GFP circuit).

Go to 7:15 of this video.

http://www.youtube.com/watch?v=ihforRdL-TU

I would say that there is at least one AFCI person that says (and shows) that a combo AFCI will interrupt a series arc.

 

[K8MHZ]

So the arc that my wife's vacuum sweeper creates every time she attempts to use it on afci protected circuits is parallel and not series. Hmmm.

Up until the point she runs it over the cord, anyway. :lol:

 

[ELA]

There was a lot written a while back about these devices ability to detect and protect against a series arc. There were all sorts of references to the 5A level as being the minimum required to detect a series arc.
There has been a lot written more recently about their in-ability to actually accomplish that feat.

Way back when I researched the QO120AFCI I was unable to get it to trip on a low level series arc ( hot to neu). I Could very readily get it it trip on the 75A parallel level.

Re: Vacuums. I was recently taking noise measurements on another device when my wife fired up the vacuum on another circuit over 50ft away from my measurement location.
Vacuums put out some nasty noise as shown in the green trace. The yellow trace is the 60hz line ( after a primary high pass filter that reduces its amplitude)

 

[templdl]

Most household breakers have a 200 amp short circuit rating, meaning for a very short period of time, they can carry 200 amps and stay closed. This is actually needed to assure that devices with large amounts of in rush currents don't trip the breaker.

QUOTE]

You have provided a very good explaination of the AFCI. Good stuff.
I would like to address the "short circuit rating" description. The basic residential MCCB has a short ciruit rating of 10kaic meaning that it has the capability of interupting up to 10ka. You reference is actually the magnetic or instantaneous pick up calibration of 200a which means that the breaker will trip instantaneously when subjected to any current above 200a. Even so if there wuld be a bolted fault that occurs the actual current would rise instantaneously to the maximum available fault current which could be as high as 10kaic which is the maximum rating of the MCCB. However most common faults are arcing in nature where currents are much less.
I'm not sure where the AFCI
As such the 200a you have refered to is the magnetic or instantaneous trip calibration. It is interesting to note that the common QO breaker's mag cal is lower and they have an optional high mag breaker where the C-h standard is high and they have an optional low mag.
I'm not sure where the AFCIs are calibrated and would have to do so research of the trip curves to determine that which is not on my priority list at this time.

 

[don_resqcapt19]

I am not sure what the instantaneous trip point of the AFCI breaker is, but the UL standard does not even require that the AFCI look at the arc signature unless the current is 75 amps or greater for 8 half cycles. Some of the manufactures look at arc signatures below the 75 amp level but the standard does not require them to do so.

 

[templdl]

I am not sure what the instantaneous trip point of the AFCI breaker is, but the UL standard does not even require that the AFCI look at the arc signature unless the current is 75 amps or greater for 8 half cycles. Some of the manufactures look at arc signatures below the 75 amp level but the standard does not require them to do so.

I looked briefly for trip curves for the C-H and Sq-D AFCI breakers and found none. It is of my opinion that the TM curve would be similar or the same as the standard breakers for each of those manufactures. Since Sq-D's standard is a low mag trip I would like to believe the C-H would have assume that they went with low mag also instead of their standard mag trip but I can't find anything in their tech data to varify that.
When looking at the standard C-H115 for example the allowable mag calibration range if between 800 and 4000A for their standard breaker and between 600 and 1600A for their low mag. All breakers withing a range of ampacity with have the same mag trip calibraton since it would be impractical to do otherwise. In other word a 15at would be the same as a 30at.

 

[ELA]

On the QO120AFCI I tested I attained Instantaneous trip at ~ 195 amps. This is an electronic trip as opposed to a magnetic.

 

[macmikeman]

I pulled this picture out of a different elec forum that was posted in response to a question I had about random tripping being possibly derived by a random number pick in firmware on purpose. I have no idea if this schematic is still the way the devices are made or if there have been fundamental changes to the design.

 

[templdl]

On the QO120AFCI I tested I attained Instantaneous trip at ~ 195 amps. This is an electronic trip as opposed to a magnetic.

Interesting. I don't think a mechanical mag trip would even come close to that lat alone being able to calibrate it for mass production purposes.
What method did you use to do your testing, ramp, or pulse?

 

[ELA]

The Instantaneous trip was a side note to other testing I was doing to Simulate an arc signature in order to learn more about how these devices worked a few years back here:
http://forums.mikeholt.com/showthread.php?t=111163&p=1016135#post1016135

As shown in the partial schematic the device has a "test Injection" coil that it uses to to test the device when you push the test button. In my testing I took over this coil and used it to inject the various frequency components required to simulate the arc. With all the required frequency components it only required ~75A of current at 60hz to produce a trip. I then removed all the high frequency components and slowly increased (ramped) only the 60hz component until the device tripped. That injected 60hz component was equal to what a line value of 195A would produce in the sense coil.
Not a highly calibrated test but should be pretty close.

 

[templdl]

The Instantaneous trip was a side note to other testing I was doing to Simulate an arc signature in order to learn more about how these devices worked a few years back here:
http://forums.mikeholt.com/showthread.php?t=111163&p=1016135#post1016135

As shown in the partial schematic the device has a "test Injection" coil that it uses to to test the device when you push the test button. In my testing I took over this coil and used it to inject the various frequency components required to simulate the arc. With all the required frequency components it only required ~75A of current at 60hz to produce a trip. I then removed all the high frequency components and slowly increased (ramped) only the 60hz component until the device tripped. That injected 60hz component was equal to what a line value of 195A would produce in the sense coil.
Not a highly calibrated test but should be pretty close.

Then your test was isolated to the coil and not the thermal element of the breaker. Commonly if the current is ramped the thermal element may trip the breaker before the magnetic element is allowed to trip. This is the issue that is commonly the result of field testing. In mad only MCPs 150a and lower they actually incorporate a wire wound solinoid which is more often than not damaged, I.e. burned up, when there is an attempt to find their tripping point by ramping the current.
It is quite interesting that you were inquisitive enough to conduct such a test though. It's fun to take devices appart to see what makes them work isn't it.
It may be an intersting point that with the instantaneous that low it could result in nuisance tripping