Electrical circuit "noise"

@cdnorris Thank you for taking the time to inform of your companies research, I appreciated it.

Is it primarily back EMF you are analyzing? Or something else? It is easy to imagine how something like an electric motor could generate a signal that could be a sign of coming failure. Brushes on an armature for example might wear and arc and this could be detected on the AC line. 

A lot of the noise we have is RFI. Or at least we think it is. I have done simple tests that would seem to indicate it is RFI.

If for example you turn off breakers going to circuits with appliances running and the sound improves (it does) that could be due to the sort of line noise your AI is designed to detect. But then if you also turn off breakers going to circuits with no appliances, no nothing even plugged in, and the sound also improves this cannot be due to back EMF or appliance noise. The only thing I can think of then is RFI. This does happen. I have tested it multiple times, including blind tests where the listener has no idea what is going on, they definitely hear the improvement. (It is not subtle!) 

So curious to know if your work is so specialized it only detects the one and not the other, or can you differentiate? Because yours is the first to be analyzing like this (patent, duh). But at the same time I would like to think that at least some of the designers of conditioners are looking into this, if for no other reason than if you want to eliminate noise it usually helps to know where it is coming from. Because in general the more you know about what you are trying to do the better your chances of actually doing it. If they are studying anything like this, I sure don't recall ever hearing about it.

@cdnorris Thank you for sharing what your company has learned. Rejection or reduction of RFI and other forms of analog EMI seems to be the improvement that many audio products seen as "snake oil" can actually make in sound quality. The 'bits are bits' crowd doesn't seem to understand that that RFI from the analog domain can generate jitter in the digital domain.

Thanks for your questions. We currently work with large commercial businesses, data centers, etc. where things like pumps, compressors, and large equipment do fail, and usually at the most inopportune times. Imagine running a lab dependent on refrigeration and having the compressors fail, or a hotel where the air conditioner goes out on a 100 degree day. As a result, our current work is primarily at the kilowatt level, which would be much higher power than what we deal with in audio. So we do not try to detect things like rfi in circuits, because it is at such a low level that it becomes noise to our processes. But the point and principles are the same - in our early years we could determine when any kind of household electrical device (like an iphone) was plugged into an outlet - residential is just not a market for us at the present time. So there is no reason that our technology cannot be used for rfi, we would probably need to amplify it (just like our sound systems do) it is just not on our current product roadmap. What seems obvious to us from our processing of electrical signals, is that there is more going on in a circuit than just a one-way movement of electrons or analog signal. As in much of life, it is complicated. Anything plugged into that circuit emits an electronic signature into the circuit, and that noise is at a level easy to detect. So it seems reasonable that it can be heard, or smear sounds or mask them, or impact soundstage, or generate jitter - and for those of us trying to listen to music in the best way we can, at whatever level we can afford, that can make a difference in what we hear, and it would seem prudent to have some strategy to deal with it.

Well said. Yours is the first I have seen of anyone objectively analyzing this down to the level of being able to detect devices on a circuit. Makes total sense. 

Probably many different things are going on all at once. AC current starts to flow, but being audio never in a continuous steady flow, so it hits any connection or device there is an impedance shift and a wave reflects back. That is just one. Then the device whether it be brushes on a stator like your air conditioner pumps or diode switching in a component it probably emits another different wave going back into the line. 

All kinds of things like this. As you say it is one thing to detect and prevent a failure, quite another to listen and analyze in order to determine which of these various noise sources does what in terms of sound quality. Then maybe we find out one sort of noise flattens the sound stage. Just guessing here, a total for instance, but if we knew that then maybe we could design something tailored or focused to do that.

To think of just one possible application.

Great stuff. If you knew the stuff some of us are using, I could hook you up with someone you could really do some great work with. Seriously.