FDC1004: FDC1004 not linear for liquid in sponge

Haydn,

Thanks for the thorough description.

Some questions:

1. Do the FDC2214 and FDC1004 use the same sensors?
2. How are the sensors configured and connected to the two devices?
3. How is the ground wire configured/connected to the tank contents?
4. What is the sensor frequency for the FDC2214?
5. What are the capacitance readings using the two devices for full & empty tanks with and without sponges?

Regards,
John

Hi John,

1/ Do the FDC2214 and FDC1004 use the same sensors?

Yes, the PCB is connected to the sensor plates with 2 wires about 20mm long.  I unsoldered the FDC1004 board, and soldered the FDC2214 board to the sensor plates at the same place. So I’ve reused the same reservoir and sensor plates.

2/ How are the sensors configured and connected to the two devices?

As mentioned, just soldered from plates to the PCB. The PCB is underneath the reservoir,

3/ How is the ground wire configured/connected to the tank contents?

There is a potential-free metal collar in the centre of the reservoir, with a coil inside that. It’s held in place by the collar and I’ve grounded the one end of the coil with a wire sticking out the top of the reservoir. The liquid is in contact with all of that, so the liquid is touching the coil = ground.

The sensor plates are on the outside of the reservoir and not touching the liquid.

4/ What is the sensor frequency for the FDC2214?

I would assume around 8MHz, but I’ll have to double check that and get back to you.

5/What are the capacitance readings using the two devices for full & empty tanks with and without sponges?

I don’t convert to capacitance, I just use the raw data values. Here are the configurations and values.

FDC2214 setup

The FDC2214 channels are all configured like this:

    channel_settings[0].RCOUNT = 20000;

    channel_settings[0].OFFSET = 0;

    channel_settings[0].SETTLECOUNT = 20000;

    channel_settings[0].FIN_SEL = 1;

    channel_settings[0].FREF_DIVIDER = 5;

    channel_settings[0].DRIVE_CURRENT = 16;

The numbers listed below are right shifted by 11bits, since those bits are very unstable for me.

FDC2214 with oil and sponge:  
0% = 122964     and 126066

100% = 121502 and 124908

fill and drain are very similar values.

FDC2214 – I haven’t tested oil without a sponge.

FDC1004 setup

configuration_registers[CONF_MEAS1] = CHA_CIN1 | CHB_CAPDAC | ((CAPDAC_MASK & capdac_value) << CAPDAC_SHIFT);

configuration_registers[CONF_MEAS2] = CHA_CIN4 | CHB_CAPDAC | ((CAPDAC_MASK & capdac_value) << CAPDAC_SHIFT);

where capdac_value =  0;

configuration_registers[CONF_MEAS3] = CHA_CIN2 | CHB_DISABLED | ((CAPDAC_MASK & 0) << CAPDAC_SHIFT);

configuration_registers[CONF_MEAS4] = CHA_CIN3 | CHB_DISABLED | ((CAPDAC_MASK & 0) << CAPDAC_SHIFT);

configuration_registers[FDC_CONF] = FDC_RATE_100 | FDC_REPEAT | FDC_MEAS_1 | FDC_MEAS_2 | FDC_MEAS_3;

Signal measurements are MEAS1 and MEAS2.

The FDC1004 values are converted to twos-complement, right shifted by 10,  and then 1000 added to them to ensure only positive integers. The numbers I’ve listed below is this final result.

FDC1004 with oil and sponge

Filling:

0%  =  3023 and 3111

100% = 3875 and 3963

Draining:

100% = 3893  and 3959

Around 16% = 3721 and 3847     (which are the values equivalent to around 80% full, on the filling sequence)

FDC1004 – I haven’t tested oil and no sponge.

FDC1004 with water and no sponge (gravity only)

Fill:

0% = 3578 and 3637

100% = 4201 and 4358

Drain:

100% = 4196 and 4363

0% = 3752 and 3750

Regards,

Hadyn

Haydn,

Thanks for the detailed info. It will take a day or two for me to go thru it. 

In the mean-time, are you using single-ended sensor measurements for both devices?
Do the results change if you don't ground the liquid under measurement?

regards,
John

Correct, both devices are setup with 2 single-ended sensor measurements.

Yes, I get a better range of values between 0% full and 100% full if I ground the liquid. 

Haydn,

Does the FDC1004 behavior change with or without the sponge if the liquid is not grounded?

Regards,
John

HI John, 

I've just tested that.  Grounded and  not-grounded, I get the same response with the sponge: ie, a wet-then-squeezed sponge reports nearly as much as a fully wet sponge.

Without the sponge (for both grounded and not-grounded) the results are roughly linear with the volume of liquid.  However, as mentioned before, grounding gives a larger range between 0% and 100%, compared to not-grounded.

H

Haydn,

Do your FDC2214 sensor connections resemble the first figure below, the second figure, or something else?

What are the inductor and capacitor values you are using on your FDC2214 PCB?
Were you able to confirm the8MHz nominal sensor frequency for the FDC2214?

Regards.
John

Hi John,

for both FDC2214 and FDC1004 implementations we have constructed, it's like Figure 54 (single-ended sensor configuration).

My tank components are  18uH (NLV25T-180J-EFD from TDK)   and 2.7pF  +- 0.5pF (RF15N2R7A500CT from Walsin)

I'm just waiting on a near-field probe to check the sensor frequency.

H

Haydn,

Just so you know, you can usually probe the sensor node using a high impedance oscope probe with a leaded 1k resistor between the probe tip and the test point. 

regards,
John

Haydn,

Everything that has been said so far makes it sound like residual oil in the sponge after draining is being detected by the FDC1004, but not the FDC2214.
The two devices use different input buffers and architectures, so it is possible for the FDC1004 to have greater sensitivity and resolution than the FDC2214 even though the data sheets report 16b and 28b respectively.

Can you share schematics and mechanical diagrams of your system with the tank, sponge and sensors?

regards,
John

Hi John,

with 1k resistor probe measurement:

the 2 signal measurement channels are

empty cavity*     :   7.1 MHz and 7.4MHz 

100% liquid full  :   6.9MHz  and  7.2MHz 

reference channel  :   8.9MHz 

NOTES

* nothing between the sensor plates, although they sit behind a 3D printed wall

100MHZ scope, 100MHz scope probe with 1k series resistor

Vpk of the measurement signals are 1.6V - 1.7V but my reference signal measurement is nearly 2.2V (I'll have to lower that.)

Hadyn

Hi John,

schematics are fairly plain 

I've added extra bulk capacitor 10uF and series 100ohm in the i2c clock to try reduce noise although it's not clear they help. I don't use the INT signal nor the SD signal.

Can I email you privately the mechanical diagrams?

Regards,

Hadyn

Haydn,

I will send you an E2E friend request that will reference this thread. Once you accept that, we can exchange info privately.

Regards,
John