LDC1612 oscillating at the incorrect frequency.

I will try to get you a scope shot tomorrow when I have everything available.

By 'correct readings' I mean that the EVM correctly reads my circuit with a total inductance value of ~5.0uH - a 4.7uH in series with a 0.3 uH.  

Here are some scope shots of the board.

This is the signal at the IN1B pin of the LDC, before the series inductor.

This is the signal after the series inductor.

The actual sensor coil (~0.3uH) is located off of the board, attached to a screw terminal by two wires about 3 inches long.  This is the signal after the series inductor, at the screw terminal before the sensor coil.

This is the signal after the sensor coil. The signal remains the same from this point back to the IN1A pin.

I connected wires from IN1A and IN1B on my LDC1612EVM to the IN1A and IN1B pins on my prototype board and tested everything in-place.  This is the signal all throughout the system on the prototype board when driven and measured by the EVM.  Everything appears to be correct, although I think there is some parasitic capacitance that I have not accounted for changing the end values slightly.

What could be causing this?  My initial thought is something to do with the drive current, although I haven't had any luck changing the settings on my prototype board.

Hello Jtee,

Please confirm that your sensor capacitor is in parallel with the series connection of two inductors. You cannot have the sensor capacitor across only the small inductor.

Regards,

ChrisO

Yes, the capacitor is in parallel with the series connection of the two inductors, just like in the diagram at this blog post: e2e.ti.com/.../inductive-sensing-how-to-use-a-tiny-2mm-pcb-inductor-as-a-sensor

All scope traces shown are taken with respect to ground.  My capacitor is definitely the correct value, 220pF, and the series inductor is definitely the correct value at ~4.7uH (give some wiggle room for part tolerances of course). 

I will try increasing the drive current beyond what I already have and see what happens.  At this point I am fairly certain that it is a setting issue and not a sensor tank issue, once again as running leads from IN1A and IN1B of the LDC1612EVM to the IN1A and IN1B pins on my prototype board yield the expected results in the EVM GUI.

I will post pictures as soon as I am able to.

Ben,

So far I have not been able to resolve this issue.  I firmly believe it to be an issue with device settings at this point. I have now created a test board containing only a microcontroller, the LDC, basic power supply components, and my LC tank.  The same issue still persists, but once again the LC tank component works flawlessly when connected to an LDC1612EVM. 

Chris,

I tried adjusting the current in increments between the minimum and maximum allowed values, and saw no change in behavior.  At this time I have been refused permission to post photos of the prototype board due to trade secret concerns in the pre-market product. 

Hi JTee,

Also, when you connected your sensor to the EVM, you said that you were getting the correct readings; can you compare the sensor frequency when the EVM drives your sensor to when your system drives the sensor?

Regards,

ChrisO

The oscillation frequencies (EVM compared to custom system) do not match.

I observe the same behaviour as JTee and have verified all which was mentioned in this thread before. My LC tank is 100 pF parallel to 55 µH. The EVM gives a realistic value for L; my system reads about 290 µH. The oscillation frequency measured in my system is roughly 900 kHz, at the EVM it is 2 MHz.

I set the EVM to the same settings as in my application. The EVM works fine with these.While applying the settings to the EVM, I noticed two things:

- The Input and Reference Divider Settings are set to 0 in the EVM by default. The LDC1612 datasheet states that "0" was reserved and should not be used.
- There is a Register called "System Clock Divider", Address 0x1D, which is not documented in the LDC1612 datasheet. The System Clock Divider is set to 2 by Default (register value 0x0200).

Changing these divider settings in my application did not help; so the EVM and my custom LDC application now run on the same settings, but read different inductance values.

When using a much larger capacitor value (4700 pF), the inductance reading in my system is "correct" (55 µH, as per design), and the EVM reads 48 µH.

Hello Ben,

Thanks for your reply. I have updated the EVM Firmware and checked the hardware and software configuration in the new GUI. The parallel capacitor value matches the capacitor which I have on my board. The GUI shows an oscillation frequency of 2.1 MHz, which is as calculated (100 pF, 57 µH). I can measure this frequency with my scope, too.

With the "larger" tank (4700 pF, 57 µH) attached, the inductance and frequency values shown in the GUI do not match the measurements with my scope or my custom system (EVM GUI reads 345 kHz, scope/custom systems read 331 kHz). Also, the GUI reading is very noisy (the value quickly changes about 5 kHz), while the signal observed on the scope is steady. This might be a glitch issue (deglitch filter set to 1 MHz), which is why I wanted to try a "smaller" tank capacitor (100 pF) in the first place.

I think I have just stumbled upon the cause for the deviations in the 100 pF case. I am using an analog switch (TS3A24159) between the LDC and the LC tank. Applications with a multiplexer are possible (as seen here). Unfortunately, the TS3A24159 has a C_ON of 224 pF at NC/NO and an additional 250 pF at COM. With 100 pF on the LC tank, this makes 600 pF; calculating with 600 pF instead of 100 pF results in much better inductance values. When I use 4700 pF at the LC tank, the capacitances of the analog multiplexer are almost neglectable. The MUX proposed in the TI Reference design is a Special High-Speed USB Multiplexer with a much lower capacitance value (7.5 pF). I will give it a try with a "better" multiplexer.

Is the LDC1612 more prone to glitches at lower oscillation frequencies?