I will rephrase the question (I reckon it was very vague) All of this happens on 120v 60hz mains. To make it simple (there are actually other loads involved), I need to turn OFF load B (12 amps) when load A turns on (12 amps), but load B need to be turned off before the circuit breaker (15A) trips. Thus my wanting to detect current flow faster than the circuit breaker will trip (~16ms). Load B (1500 watts grill) is considered ON all the time (thermostat controlled), and load A turns on at random moments (1500 watts coffee machine).

I never want the circuit breaker to see more than 15A at any given time.

  • 2
    \$\begingroup\$ What's the reason for wanting the fast detection in the first place, human safety or something else? It might be good to describe the overall results you're after. \$\endgroup\$
    – PeterJ
    Jun 1, 2013 at 6:13
  • 1
    \$\begingroup\$ What you ask can't be done. Every measurement will have some error or other noise, so detecting exactly zero versus non-zero is physically impossible. Also, you ask for really fast, then say anything less than 15 ms is acceptable. This is really slow in a lot of contexts, which is why we don't use handwaving like that in engineering. Step back two levels and explain what you are really trying to accomplish. What voltage and frequency is this AC you want to measure. What is the real current detect threshold, and no, "ANY" is not a meaningful value. \$\endgroup\$ Jun 1, 2013 at 13:26
  • \$\begingroup\$ To answer this I'd need to know what the max current will be ( If your maximum current is 100A you may need some thing different compared to 10mA maximum current). What is the threshold for indicating current is flowing? 10 uA or 10 mA ? .Can you insert something in the current path. Or can it be by capacitance or inductance? Let me guess a few values to get you started. Maximum current 10A, trigger level is 1mA, must be non contact. Circuit Maximum voltage 24VDC. What did I guess wrong? (Yes, I have faced the Design Engineer who wouldn't give me a spec too often, so I guess / estimate) \$\endgroup\$
    – Spoon
    Jun 1, 2013 at 23:11
  • \$\begingroup\$ Circuit breakers take time to trigger, typically a few seconds for light overloads, to several tens to hundreds of milliseconds during extreme overload. Allowing an overlap of a AC cycle or two will very likely not trip the breaker anyways. \$\endgroup\$ Jun 3, 2013 at 14:13

2 Answers 2


You cannot definitively sense AC current in a period of less then ~16 mS, as a device may not draw power over the entire AC waveform. As such, you need to observe the current draw over an entire AC cycle before you can be certain there is no power draw.

The cycle period is \$\frac{1}{60} = 0.0166...\$ Seconds, or 16 milliseconds.

This is for 60 Hz mains. 50 Hz mains is: \$\frac{1}{50} = 0.02\$ seconds, or 20 mS.

Note that if the device you are sensing the current draw from draws current like a resistive load, you may be able to detect it's presence more rapidly. However, you cannot definitively say there is no device present until an entire cycle has elapsed.

  • \$\begingroup\$ Actually just a question for discussion. Could a frequency multiplier, supposing you have a device that can do that in this situation, make the measurement more rapidly? \$\endgroup\$
    – jhc
    Jun 1, 2013 at 6:49
  • \$\begingroup\$ @jhc - Only if you're powering the actual device under load with the multiplied frequency. Remember, it's not your device that is producing the 60/50 Hz, it's the power plant. \$\endgroup\$ Jun 1, 2013 at 7:03
  • \$\begingroup\$ Airplanes generally use 400/440Hz power, so if your appliance supports that, you could use a VFD to synthesize 400 Hz mains, and then monitor the current draw on that, but that would be a LOT of work. If you're going through that much effort, it would probably be easier to simply modify the device being measured to output whether it's powered. \$\endgroup\$ Jun 1, 2013 at 7:04
  • \$\begingroup\$ Well, I imagine it would be possible to put a parallel load and apply frequency multiplication there. The power source would have to be able to handle these harmonics though. EDIT - Probably won't work, parallel load will force the higher frequency voltage onto device, who knows what will happen then. \$\endgroup\$
    – jhc
    Jun 1, 2013 at 8:03
  • \$\begingroup\$ You are making the assumption that "AC" means 60 Hz line power. I don't see that anywhere in the question. If you are going to assume additional criteria, you should at least say so clearly. For example, even sticking to line power, your 16 ms number would be 20 ms in many parts of the world. But we still don't know if the OP is trying to detect audio, for example, or something else. You really can't answer the question in its current form. \$\endgroup\$ Jun 1, 2013 at 13:29

The fastest current sensor on the market is a low inductance coaxial current shunt. They go from DC to 2GHz bandwidth, depending on the resistance value.

LEM current sensors are also very fast. Their BW goes up to 800kHz.

With these current sensors you are pretty much limited by your current detection circuitry: thresholds you use and delays that you introduce with comparators.


Not the answer you're looking for? Browse other questions tagged or ask your own question.