Hello Team,
In webench coil design, it shows distance resolution.
Below first row LDC coiil shows distance resolution of 0.0207um.
May I know if 0.0207um is the tested performance that LDC1612 system can achieve?
Since there is no LDC1612 noise floor specified in datasheet, I am not sure if the 0.0207um includes LDC1612 noise floor.
Thank you.
Regards,
Ting
Hello Ting,
This kind of resolution is likely not practical especially if this is an absolute proximity measurement.
From our experience, the LDC161x family does have very good resolution to detect relative changes in metal proximity but a number of external factors must be well controlled and additional averaging may be required. For example, take a look at the front page of the LDC1612 datasheet which shows data collected during characterization which shows well below 0.1µm resolution.
Therefore, it is likely that the webench tool is making the calculations for a best case scenario, but it cannot account for all of the potential noise sources, such as reference clock quality, sampling time, number of averages allowed, proximity to nearby metals, sensor parameters such as Q factor with metal present, temperature, etc.
This is why there is a difference between resolution, precision, and accuracy. It may be possible to calculate or correlate a delta in frequency to a delta in distance (resolution), but if there is noise sources that make the repeatability of the measurement change from measurement to measurement (precision) then your expected correlation to the actual distance (accuracy) goes down. In practice, it is possible to get very good accuracy (~1µm) with reasonable system parameters for relative measurement applications with the LDC161x family. Good examples include Metal Touch buttons such as the 16 button stainless steel keypad where the metal is at a fixed distance away from the sensor and we are able to detect when a user slightly presses the metal surface which causes a slight deflection (<1µm) towards the sensor. The key is that the metal-to-sensor spacing is <20% of the coil diameter away (maximizing the sensitivity) and the absolute distance reading is not needed as an output. We are only detecting the relative changes in metal proximity caused by the button press.
Ideally, you could enter everything into webench and it could directly calculate the real world performance, but the tool is not that advanced at the moment. In general, it is best to build up the solution and test the true performance and we've developed a number of prototyping tools (LDCCOILEVM, LDCTOUCHCOMCOILEVM, LDC2114EVM) and design collateral (www.ti.com/.../inductive-sensing-technical-documents.page) to help with the prototyping phase.
Regards,
Luke
