Increasing accuracy in the collection of data coming from sensors is a need that, sooner or later, Makers need to face. Paul Martinsen from MegunoLink created a tutorial to eliminate noise from sensor readings on Arduino with three simple filtering techniques.

The Averaging and Running Average techniques are easy to implement as they work by adding a number of measurements together, then dividing the total by the number of measurements. In both cases, the downside is that it can use a lot of memory.

The Exponential filter is a better solution for several reasons: it doesn’t require much memory, you can control how much filtering is applied with a single parameter, and it saves battery power because you don’t need to make many measurements at once. For this solution, they developed an Arduino filter library so you don’t need to go mad with math!

Interested? You can find the tutorial and explore the code on MegunoLing’s blog post here.

Network Analyzers are frequently used for measuring filters, making them extremely valuable for building radios and general mucking about with RF. They are, however, extremely expensive. You can, however, build one in an Altoids tin with an Arduino Nano, a small screen, and an AD9850 frequency synthesis module picked up on eBay.

The basic idea behind a network analyzer is to feed a frequency into a device, and measure the amplitude coming out of the device, and plot this relationship over a frequency. [Bill Meara] has been a human network analyzer before, changing frequencies and plotting the output of devices under test by hand. [DuWayne] (KV4QB) build a device to automate the entire process.

The block diagram is easy enough – an AD9850 sends a signal to the device, and this is measured by the Arduino with a small amplifier. The signal is measured again when it comes back from the device under test, and all this is plotted on a small display. Simple, and [DuWayne] is getting some very good readings with a lowpass filter and crystal filter made on a small solderless breadboard.