Yoctopuce offers a number of products which can measure alternating current, including measures on the mains. To achieve sufficient accuracy in this case, the older modules apply a sliding average over several seconds. This week, we're releasing a firmware enhancement for these modules to prevent averaging from slowing down the detection of power on events.
In a previous post, we mentioned the sampling frequency limitations of the first-generation of the Yocto-Watt. These limitations also apply to the Yocto-Volt and the Yocto-Amp: when measuring alternating current, to estimate RMS voltage and RMS current, we have chosen to apply a sliding average over a little more than one second, in order to obtain a sufficiently accurate and stable measure.
Yocto-Volt and Yocto-Amp
While this method has the advantage of accuracy for a voltage that varies little, it has the disadvantage of causing a latency when trying to detect a voltage change. Here's an example of measures made using this method, on a scale of tenths of a second:
You can see that the sliding average is clearly detrimental to power switch detection, introducing a significant delay.
So this week we modified the way the sliding average is calculated, by aligning the averaging window to the instant when the AC voltage appears. As can be seen below, this greatly reduces the latency effect:
The result is logically not as sharp and precise as with the Yocto-Watt-V2-C, which has a sampling rate ten times faster, but it's already a good improvement.
Bear in mind, however, that this modification only improves latency in cases of pure AC voltage switching. If the voltage only varies but does not disappear, the averaging is still clearly visible compared to the instantaneous measures that a Yocto-Watt-V2-C can provide:
In this case, the solution is clearly to use to a Yocto-Watt-V2-C instead: it's a little more expensive, but not without reason...
