Electromagnetic compatibility for CapTIvate user interfaces, introduction to CapTIvate peripheral features. In this brief video, we will introduce the important features in the CapTIvate peripheral that improve noise immunity.

Let's look at some of the peripheral features in CapTIvate for dealing with conducted noise, EFT, and ESD events. First, CapTIvate has an integrator-based charge transfer measurement engine. In addition, it has parasitic capacitance offset subtraction capability.

The offset subtraction allows for CapTIvate to compensate for very large parasitic capacitances in a PCB layout. Ground is very ideal in a PCB layout for improving noise immunity of a system. We'll look at this in the upcoming layout sections.

CapTIvate also has a frequency hopping oscillator with spread-spectrum clock modulation. This means that, if noise is present at one of the conversion frequencies, we can move the conversion around in frequency to auxiliary conversion frequencies and look for correlation between measurement points to determine if a touch was actually present. CapTIvate also has a dedicated, on-chip, low dropout regulator, which powers the sensitive analog measurement blocks. This provides further power conditioning to transient events on DVCC versus ground.

Here, we have a 32-button, mutual capacitance test panel from the capacitive touch EMC reference design featuring CapTIvate microcontrollers. This reference design here is being stressed to a conducted noise test of 3 volts RMS at the 4-megahertz conversion frequency that the sensors are being measured.

If we look at the test data on the bottom half of the plot, you'll notice the raw data completely unprocessed coming out of the CapTIvate peripheral versus time. Two touches were performed during this test. If you look at the bottom plot, you'll notice that we're showing the raw data from four different conversion frequencies-- the base conversion frequency being the 4 megahertz, which is shown in yellow.

Note how the base conversion frequency is heavily corrupted by the noise that's being injected at 4 megahertz and at 3 volts RMS. However, the remaining frequencies-- F1, F2, and F3-- are not affected by the noise. Because of this, we can correlate between the four different frequencies to resolve the measurement back to usable information as shown in the top plot.

In the top plot, the blue represents the filtered count, which is the processed information of all four frequencies shown in the bottom plot, as well as the black long-term average, which is our reference point where we're looking for a touch deviation from, and the green threshold marker. Note that when the touches are not present, the filtered count remains close to the long-term average, and there's no unusual deviations. During the touch, the filtered count remains clearly above the touch threshold until the touch is release.

Let's look at the same thing versus frequency in addition to just time. Here, we have that same board, again the 32-button, mutual capacitance test panel, being tested with noise frequencies sweeping from 300 kilohertz out to 80 megahertz. This shows how the frequency hopping algorithm becomes effective.

On the left hand side of the screen, we're performing a 20-volt RMS conducted noise immunity test, again sweeping 300 kilohertz out to 80 megahertz in 1% steps. On the bottom half, you can see how the raw data is corrupted at the various frequencies and their harmonics. The four fundamental conversion frequencies can be quite easily observed as the first four noise peaks in the raw data. Noise can also be seen at the harmonics out to the right.

Note that there is significant spread between the noise bands. In other words, when noise is present at F3, F0, F1, and F2 are all clear. And then there's some gap before we again start to see noise at the next frequency up, F2, at which point frequencies F0, F1, and F3 are now clear. This allows us to correlate, even at 20 volts RMS, back to the clean data shown above. The filtered count matches the LTA throughout the duration of the test with no jitter in the measurement whatsoever.

On the right hand side is a touched sweep with 10 volts RMS of conducted noise entering into the test module. In this example here, you can notice that the amount of noise coupled into the system is much more significant than in the case where no touch was present. This comes back to the example circuit we looked at previously, where the addition of the user's touch capacitance, or the additional coupling path back to earth ground, becomes a noise coupling path itself.

Note that the frequency hopping method used in the other test works equally as well in this case despite the higher noise levels that are present. There's enough separation between frequency bands to accurately create the filtered count and compare it to the threshold.

Thanks for watching this video. After watching, you should have a general idea of how the different features built into the CapTIvate peripheral can improve the robustness of a CapTIvate user interface in the presence of noise. For more information on the design process for noise immunity, watch the video on the EMC design process in this series.