My understanding is that basically what is done is to create a lot of arc fault conditions in the lab, and look at what the electrical waveform looks like at the breaker. This is done for all kinds of different cases (but obviously not all of them). Then they look at all kinds of "normal" loads, such as CFLs, refrigerators, vacuum cleaners, hair dryers, etc., to see what the electrical waveform looks like at the breaker. Then they write software that tries to analyze the electrical waveform at a breaker and detect when there is one that looks like what an arc fault creates, but not one that looks like what a "normal" load creates. If one is found that looks like an arc fault, and not "normal", the software trips the breaker.
This is a problem that is very similar to that in other fields, such as monitoring heart beats for irregularities. However, this may be one of the most difficult scenarios given the space and cost restrictions (compare the cost and size of a breaker vs. a heart monitor), and well as the similarity of the arc fault "fingerprint" to many "normal" ones. I'm not an expert, but from what I've read I'm not convinced (as apparently the decision makers have been) that they have gotten the recipe right yet. Though as time goes on and the processing power that can be put into a breaker at a given cost increases, I'd expect things to get better (as well as appliance manufactures having to design them so that they won't trip the arc faults).
So why not develop a programming module and be able to program these devices to recognize an acceptable load? At least those of us who service them will be able to look like we know what we are doing with these things. Many of us run into a problem load that the device does not like and end up telling the customer that there is nothing we can do about it. That is not good for the reputation of the installer or the equipment, but makes the handiman that solves the problem by putting in a standard breaker look like a genius to the homeowner.
I don't know how the people pushing the use of these (which I think is good, but they definately are still in infancy) ever expect them to discriminate between all possible loads that could be placed on them and known problem load conditions.
I see no reason why any manufacturer should have to redesign their equipment so that the AFCIs will work correctly. It is my opinion the issue is up to the AFCI manufacturers to solve.
My opinion also. I think they also should have been perfected more before being put into code, instead all that was important was the fact that several manufacturers needed to have one before forcing us to use them.
.. Two choices - fight the government and try to get AFCIs removed from the code, or try and change the vacuum so it doesn't trip AFCIs .
Where does the government come into play in requiring AFCI's other than in cases where government operated AHJ's adopt and enforce the NEC as it is printed? They do not make the NEC, sometimes they do amend it for whatever reason they feel is necessary. I can give you one example of government eliminating AFCI requirements:
Nebraska State Electrical Board amended out all AFCI requirements in past codes because (not an official statement but my understanding) they felt that the AFCI's were pushed into the code before they were ready for consumers to use them. The manufacturers had a lot invested in R&D and did not want to wait another code cycle or two to further perfect the product so they pushed hard to get it in code and made the consumer be the testing lab. I think Nebraska was one of very few places that completely amended all AFCI requirements to non existance in their code.
In 2008 Nebraska Electrical Division decided they wanted to adopt 2008 NEC without AFCI amendments - (maybe, I don't know all facts of why) they decided the first generation AFCI's were being replaced with supposably an improved product and now maybe we should allow them.
Problem is every three years when the new NEC gets adopted, because it will become state law, has to be passed by the state legislature. Normally this happens with little or no debate, law is introduced and passed
But the 2008 NEC and no more amendments deleting the AFCI requirements, plus the tamper resistant receptacle requirements, weather resistant receptacle requirements, had some homebuilder associations presenting some strong opposition so it got tabled and was not brought up again before legislative session was over for the year.
Same thing pretty much happened in 2009. Eventually 2008 code was adopted in middle of 2010. I'm not sure of issues involved that got it passed. I don't know if 2011 code is expected to have any similar issues or not. I'm sure a bill will be introduced to adopt 2011 and likely will not even come to the floor until maybe March, if passed will still be 90 days before it is a law unless there is an emergency clause - I doubt that is even slightly possible with that particular bill. The state budget will be getting the most attention in this years session.
