Getting Started with the LDC1000 - Small Metal Detection!

The LDC1000LDC1000 is a relatively recent device introduced by Texas Instruments about a year or two ago – it is designed to measure inductance to a high resolution (24-bit!) in an extremely convenient single-integrated-circuit solution.

Metal objects close to a coil can affect the coil’s inductance sufficiently for it to be measured with the LDC1000.

Here is TI’s LDC1000 evaluation boardLDC1000 evaluation board:

The board is a convenient USB stick with snap-able sections to separate it into the USB interface, the LDC1000 itself, and a PCB coil (approximately 19.5uH inductance).

24-bit resolution is quite granular! I was curious exactly what could be detected with a subtle change in inductance.

As a first experiment (Not the first experiment to use a clock ) I taped it onto a clock face:

Could it detect the thin second-hand as it came around every minute? The clock's second-hand was about 1.5mm in width, and stamped out of some thin sheet metal.

The aim was to see if it could cause a change in inductance significant enough to be detected despite the second-hand being several millimetres distant on the other side of the clock’s plastic face cover of course.

I ran the excellent TI-supplied software which logs into a CSV file (which can be opened using Excel), and logged for about three minutes. (The screenshot below is not from the clock experiment).

Excel isn’t fantastic with large sensor data files (in fact it is terrible – it is not designed for it), so I brought the CSV file into MATLAB. The x-axis shows the sample number (the microcontroller on the USB stick was set to sample at approximately 1000 times a second), and the y-axis shows the measured inductance. You can see that the granularity is at the limit, but nevertheless we clearly see when the second hand sweeps past every minute!

Some two lines of signal processing could be used to filter the very low noise.

coeff=ones(1, 100)/100;
ind_filtered=filter(coeff, 1, ind);

Now we have a very clear signal that indicates when the second-hand is sweeping past! The red line shows an example threshold that could be used.

The LDC1000 is extremely sensitive to inductance changes, and appears to have an extremely low noise output. Interestingly the noise does not seem to be characterised in the data sheet (presumably due to the extreme difficulty) but the automotive qualified part does have an example histogram. But the initial practical experiment shows that the noise is miniscule – truly incredible. The applications for this part are immense.

In summary, this was just a short blog post highlighting the sensitivity and low noise output from the device. Further blog posts will investigate the LDC1000 further.

Also, as an excellent resource, check out the experiments by jw0752 in his LDC1000 discussion here.