This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

FDC2214: Confirmation of "driven" and "input" pin for the sensor inputs

Part Number: FDC2214
Other Parts Discussed in Thread: FDC1004

I have an application using CH0 and CH1 for sensing liquid , and CH2 for a reference sensor. All 3 channels are configured with differential sensor plates.  CH0 and CH2 are behaving as expected, but CH1 has some strange behaviour where the readings drift and recover for no apparent reason.

Since CH0 and CH1 are measuring the same vessel of fixed volume of liquid , I doubt it's due to that.  The LC tanks are all the same (18uH and 2.7pF)  and sensor plates same size.   I wondered if the layout and specifically the proximity of ground planes and other signals to the sensor tracks are the cause.

1/ Should the channel signals be "isolated" from other circuit signals by putting a ground plane underneath and along side the channel signals?  If so , how close or how far ? Currently I have minimum track-track clearance between ground plane and the one channel signal.

2/ Please can you confirm which pin (either  CHxA  or CHxB) is the driven output, and the other pin presumably is the input?  (I'm assuming it's similar to a crystal oscillator circuit. ) Currently IN0A is routed next to ground plane and other signals,  while IN1B is routed close to ground plane and other signals.

 

U100 is the FDC2214 ;  CH0 has L100//C102 and tracks going to the left,  and CH1 has L101//C103 and tracks going to the right.

3/ I would imagine I should put the driven signal next to ground plane and other signals, and the "input" signal away from ground plane/other signals.

Thank you,

Hadyn

  • Haydn,

    I will look over the layout image you have provided and update this thread by COB tomorrow.

    Regards,
    John

  • Haydn,

    Could you provide a schematic of your circuit?

    Also, just so you know, we advise using the  FDC1004 over the FDC2xxx devices.
    Many of our customers have struggled to get to production with the FDC2xx family.

    The FDC1004 uses a switch-cap front end as opposed to a resonant circuit.
    It also has active shield drivers which can really help manage EMI, especially with liquid-level sensing applications.
    You can find a comprehensive list of FDC1004 collateral at the E2E FDC1004 FAQ page.
    Several of the app notes are oriented towards liquid level sensing.

    Regards,
    John

  • Hi John,

    Thanks for your comments.  Let me discuss internally, but here is the schematics.  Powered from USB 5V through a 3v3 LDO.

    The interrupt and SD pins are not used nor connected to CPU. I've added series resistances to i2c to try reduce i2c noise when polling for data.

    Plate 5+6  are copper pour on the PCB top and bottom layers - used as a reference compensation.

    Regards,

    Hadyn

  • Thx Haydn.

    What is sensor frequency when measured with a high-impedance probe?

    Regards,
    John

  • Hi John,

    We don't have a high-impedance probe , but I've configured the FDC2214 device for continuous conversion on single channel only (as well as shutdown to confirm the radiated emissions is  indeed from that channel ), and by sniffing nearby with 1:1 oscilloscope probe, I can see these frequencies:

    Sensor Freq (MHz) Data >> 11
    CH0 8.14 117533
    CH1 8.09 116451
    CH2 8.86 128651

    NOTES:

    Measured over 10 oscillation cycles to reduce some of the measurement error.

    Data shifted down to remove lower bits, that are un-usable.

    Normally I run it in AutoScan sequence for CH0, CH1, CH2.

    Is that sufficiently accurate?

    Regards,

    Hadyn

  • Thanks Hadyn.
    With the nominal frequencies, inductance and capacitance given here, it looks like the sensors should have about 20pF of capacitance?
    Does this line up with what you expect?

    Interestingly, there is another FDC2112 E2E thread where the sensors are in close proximity, and they are having some issues as well, which may be related to your original question (1).

    In general, capacitive sensors can be a little finicky to PCB routing and parasitic coupling between traces. 
    If you can, please try experimenting with some ground plane over and/or between your sensor traces.
    You can't go wrong with more shielding as long as it doesn't cause too much parasitic capacitance.

    Another experiment to try is along the lines of a suggestion in the other thread: different sensor frequencies for CH0 and CH1.
    If you decide to try this approach, then please make them non-harmonically related.

    Regards,
    John

  • Thanks for the response John .  I'll have a look again at the FDC1004.  Just last question, can you confirm if it's the INxA pin that is the output pin in the FDC2xxx chips?

    thanks,

    HAdyn

  • Haydn,

    The INxA and INxB pins are both outputs.
    They should both have half-cycle sinusoids displaced a period from one another.
    so if you probed them differentially you should see a full cycle sinusoid.

    Regards,
    John