Accuracy improvement for some devices

Accuracy improvement for some devices

We recently took some time to study whether it was possible to improve the accuracy of some of our analog sensors with simple software changes. In particular, we suspected that it was possible to improve the sampling strategy to reduce the noise margin on some measures. Good news: it works and a simple firmware update on some modules is sufficient to make them really more accurate on some measures :-)

The modules for which we really hoped to improve the accuracy are those used to measure the AC mains: the Yocto-Volt, the Yocto-Amp, and the Yocto-Watt. Indeed, although these modules have a 0.5% accuracy on direct current, up till now we had to limit the announced absolute accuracy at 3% for alternative current, because of the noise margin.

By managing more precisely the sampling period, and by optimizing it for expected frequencies on the mains (specifically 50Hz or 60Hz), we have been able to considerably reduce the measuring noise on the mains. We now announce an absolute accuracy of 1% on alternative current, and the sensitivity is even improved by a factor ten, which enables us to now display a decimal on the mains voltage. Here are two measures performed in identical conditions with the same Yocto-Watt, first with the old firmware, then with the new one:

Measures with the old firmware Measures with the new firmware
Decrease of measuring noise on alternative current

Furthermore, we were recently told that the Yocto-Thermocouple, that we have been selling for many years, produces noisier measures than other products on the market when in presence of low frequency electromagnetic noise (motors, and so on). We already documented in a previous post the noise which may result from ground loops and how to avoid it, but knowing that this type of noise probably included a component linked to the mains frequency, we tried to apply the same sampling strategy to this module. And here as well, this reduced the noise considerably, to the point of almost eliminating the oscillations caused by ground currents. The later however contributes to the introduction of a slight measuring bias, but the measure becomes much more exploitable:

Measures with the old firmware Measures with the new firmware
Decrease of measuring noise in the presence of ground loops

So, if you have a Yocto-Volt, a Yocto-Amp, a Yocto-Watt, or a Yocto-Thermocouple, don't hesitate to perform a small firmware update...

PS. And a big thank you to Marko Luther for the very quick turn-around on validating the Yocto-Thermocouple improvements on the Artisan software using his coffee roasting environment...

Add a comment No comment yet Back to blog

Yoctopuce, get your stuff connected.