Hello Community,
we are designing a sensor to measure the flexing of a circular stainless steel membrane under changing pressure conditions in a closed container.
The membrane is of circular shape with a diameter of 55 mm. The membrane will flex according to the pressure difference inside and outside of the container (in the container the pressure will be varying between 0.1 mbar and 1000 mbar).
The maximum distance in the center of the membrane from the sensor coil will be about 0.9 mm the minimum distance will be 0.5 mm.
So our goal is to maximize the resolution of the measurement for the delta of 0.4 mm between both membrane distances.
The movement of the target (the stainless steel membrane) is orthogonal to the sensor plane. The PCB Coil and the membrane are positioned centered above each other.
The measurement frequency does not really matter for us, we would be fine with 10 Hz or even less since the application is more long-term monitoring the pressure curves with alarms being set when defined thresholds are exceeded.
We did a smaller scale prototype with a 25mm membrane and a 14mm coil on the PCB (single layer coil on the side of the PCV that faces the membrane) with 18 Turns and a calculated oscillation frequency according to the we-bench tool of 5 Mhz (with a 400pF capacitor). The prototype showed that it is absolutely a working solution yet we would like to increase the resolution as much as possible in general and also perform everything possible to compensate for temperature drifts. The planned temperature range for the sensor that might be exposed to external conditions could be anything from -10 to +80 degrees °C.
While digging through the excellent application notes for LDC sensors we already gathered a lot of input, yet we still could not find the right answer in regards to coils size when you only want to measure a rather narrow band of distances.
Application notes say to chose the largest possible outer diameter of the coil and keep the inner diameter also rather large (ratio dinner / douter> 0.3) but then there are also some excuses from the rule where a smaller inner diameter seems to provide a higher resolution (as an example close proximity metal-touch-button applications are mentioned but without further details).
So we could increase the outer diameter to something like 40mm if needed, yet we actually want to focus on the movement of the center of the membrane since the change in distance will be the highest there (the closer we come to the border of the membrane, the less (orthogonal) movement / change in distance we will have).
So here are finally my questions:
1.) what is roughly the recommended outer and inner diameter for the PCB Coil when we want to measure distances from 0.5 mm to 0.9 mm only? Is it beneficial in general to build the sensor in a way that the minimum expected distance to the target is as low as possible (in our case 0.5mm) or would it be better to increase the minimum distance a bit?
2.) from the application notes I figured that we should use an external Clock Signal (40 Mhz for two channels/35Mhz for a sig) rather than the internal Clock generator and try to reach a max coil frequency of about 3-5 Mhz to still keep a somewhat high sensor resolution (high amount of individual output values / steps)
3.) Should we go for a single PCB Coil (on the PCB layer facing the target) or would it make sense to use a two-layer coil (bottom and top layer of the PCB) instead?
3.1) Also when talking about temperature compensation it seems more suitable to only use a single layer coil to eliminate coil distance changes due to temperature changes
4.) In general, I checked the Application Report SNAA212A (LDC100x Temperature Compensation) for input on the reduction of temperature drifts, my "conclusion" was that the following should be considered:
- use a single-layer coil to eliminate the influence of the PCB expansion over a wide temperature range
- use an external clock signal
- use capacitors with a low-temperature variation (C0G/NP0 capacitors)
- use the same material for Coil and target (not an option for us, we have to stick to a stainless steel membrane)
- the distance from the coil to the target seems to be relevant in general too (quote from AR "LDC100x Temperature Compensation"):
-
When a target is in proximity of the coil (<50% of the coil diameter distance), temperature effects on the mutual inductance need to be evaluated. A temperature variation changes resistivity, and consequently eddy current distribution in the target. This change in eddy current distribution impacts mutual inductance. The magnitude of the impact depends greatly on the distance to target as well as frequency, and is on the order of tens of ppm when the target is very close to the coil, quickly dropping to single-digit ppm when the target is at a distance greater than 20% of the coil diameter.
- So it seems we should try to adapt the design to ensure that we have a minimum distance between target and coil of 20% of the outer diameter. Currently, we have a 0.5 mm minimum distance but it should rather be 4mm for a 18mm coil if I get this right, which might not be an option for us since we need to build a rather flat sensor. We can probably go to a minimum distance of 2mm, but we are not sure if this actually negatively impacts the resolution of the sensor (the max delta between the highest and lowest distance would still be 0.4mm)
- Also, the frequency should be rather lower than higher in general. According to the AR 1 Mhz has a much lower temperature drift than e.g. 5 Mhz
- On the other hand, due to the target material being stainless steel, we cannot go to low in the frequency to still get a suitable skin depth and general sensibility. I would currently aim for a frequency of maybe 3 Mhz as a compromise
- Implement a mapping table in software in combination with a temperature sensor on the same PCB to compensate for temperature drift (this would of course be the last resort, but is certainly doable)
Are there any other recommendations other than that from your experience to reduce the influence of temperature on the measurement?
Ok, wall of text, thanks to anyone who made it this far :-) Any reply is highly appreciated.
Kind regards
Paul
