FDC2214: Active Shield

This sensor is working by frequency shift, it means we will get change on the reading value by the frequency shift, right??

But is it okay, if the signal amplitude is different between the active shield and the excitation signal? Will it affect the reading value?

Hello Specia,

Thanks for considering to use Texas Instruments.  Can elaborate on how you connected the cable, is it to a buffered output of the INxA channel? If you are using the active shielding with a buffer, can you provide a schematic with the buffer amplifier and FDC2214? Also where are you probing on the shield, can you provide a picture or diagram?  Are you probing directly with on oscilloscope probe?  It might help to solder a 1k resistor to the shield and then probe the shield signal from the resistor lead not connected to the shield with a high impedance probe.  The capacitance from the probe can influence the measured signal if it is directly connected to the shield.

That's what we got, even we already put 1k2 between the probe and the shield pad. You could see that blue line (2nd-probe) shows decreased amplitude

just notice the blue signal (i didn't probe the yellow one properly), is it okay if the excitation signal appeared like this? I got this since use a current drive at 867uA. As far as i know, the signal should be half-sineWave..

related to the waveform,  is it okay if the excitation signal appeared like full-sinewave?

Hey Specia,

The shield should mimic the LC tank signal.  If you do a single ended scope measurement of the LC tank you should see the kind of behavior you are observing.  Whereas if you do a differential measurement, you should see a signal that looks more like a sinusoid.  Here is what I observe from our device on our EVM.

Single ended:

Differential:

Hi Patrick,

1. Is it okay if the excitation signal appeared like this? Similar to half-sine-wave, but not perfect when I used a current drive at 0.867mA. Is it acceptable by the sensor? The configuration type of my application is single-ended. I got  perfect half-sine-wave like you when the current drive at 0.169mA.

2. How far is the effect of tuning the current drive? Should i tune the current drive until get perfect half-sine-wave?

Thanks

Hey Specia,

1- Depending on the channels used and current setting you could drive the excitation signal with up to 1.5mA or possibly higher, with the only stated max being 8mA in abs max table.  So I think from a current standpoint its fine.

2-While the design could benefit from a cleaner sine-wave, I think adjusting the current drive so that there is some margin above the minimum voltage amplitude specification of 1.2V is more necessary.

still about this sensor, is it okay to connect FDC2214 to analog multiplexer? any suggestions for the analog multiplexer?

Hello Specia,

I think it might be challenging to implement.  The multiplexer will likely have some affects on LC inductance and capacitance and might introduce noise thereby complicating the capacitor plate design.  Also when acquiring data during multiplexing between capacitors, you may need to discard the first set of results after switching multiplexer channels.  Presently I am not aware of a reference design that does this, and I am unable to suggest a suitable analog multiplexer.

Hi Patrick,

This is when i tried to connect the sensor to an analog mux. Don't you think it's like harmonic with something else? Are you noticing something from this picture?

Hello Specia,

Are you referring to the picture you posted a few days prior?  That does look like a half sinusoid with at least another frequency element mixed into it.  The repeatability of it does suggest its not Gaussian noise and that it might be a higher order harmonic. I think you could probably verify with a spectrum analyzer.  Where is the yellow and cyan signal in relation to the mux?

Hi Patrick, pardon. I mean about this picture while i am connecting the analog mux. I just used channel 0 only of the sensor.

 and this is the sensor reading after I coded the sensor using Python. You could see that only negative spikes existed while I was connecting the analog mux. Are you noticing something from this picture? If the negative spike doesn't exist, i guess the noise will be low.

Hello Specia,

The graph you are showing above looks noisier than what I see from running the device with our EVM.  It looks like the mux you are using is contributing noise to the signal.

Hi Patrick, is the negative spike issue related to crosstalk or charge injection of the analog mux? I've tried using an electromechanical relay, but the issue has not appeared yet.

Hello Specia,

It could be either.  Have you tried reaching out to the mux provider?  They might be able to provide better insight into how the mux is loading our device and the LC tank.

Hi Patrick,

Alright, currently i am trying to ask them as well for this issue.

Btw, i have another question, what do you think if the excitation signal (peak) has amplitude >1.8 V?

I saw in the datasheet, it said the max amplitude is typically 1.8V.

You could see i obtained more than 1.8V for the amplitude.

what is the meaning of Vpk? 

It should be like that, right? And so it should be between 1.2 - 1.8 V?

And this is the signal when I set the current drive to minimum at 0.016 mA.

Should I change the LC-tank to fit the oscillation amplitude so it can reach between 1.2-1.8V?

I also have other question, how to disable the sensor for a while? Could I use the internal mux to disconnect the INA and INB pins?

Yes, Vpk should be as you indicated in your drawing and within 1.2 and 1.8V. Per the datasheet I would try to stay within those bounds.  I will see if I can get some input from our designers as to why. If you switch channels internally, then only the active channel should be influencing the measurements you are reading from the device.

Hi Patrick, 

Currently i am trying to make it to the optimal range (1.2-1.8Vpk).

I found something interesting here, do you the meaning of filter capacitors they said? Regarding to the spikes issue.

I tried to find the filter capacitor at INA and INB connection in the FDC2214EVM schematic, but I didn't find it, cmiiw

are they the DNP capacitor?

Hey Specia,

Below are the recommended capacitor values per the datasheet.  These are all on the bottom side of the evm.  Its possible these were all not populated on the shipped EVM you have.

As for the choice of values used, its not clear how the original author chose those values.  However, I suspect this resource might help you fine tune those values for your case.

Hi Patrick, i am using single-ended mode for this sensor. For example I will use the top channel only, so I have to populate C11, C12 right? I don't need to do for C16, C18

Hello Specia,

I think if you are installing C11 then you should also install C16, likewise if you install C12, you should also install C18.  Single-ended appears to relate to the sensing capacitor configuration and not the generation of the signal and the impact of external emissions.  Per the datasheet, the same bit setting used for single-ended setting in the clock dividers register can also be used for a differential setting.  Additionally, I do not see anything in the configuration register's that differentiates single-ended from differential.  Therefore a differential signal will be generated in either case and therefore I think you should use a symmetric differential filter as shown above for best results.  This document in section 2.2 gives some insight on the differential filter.

Hi Patrick

This is after I added the filter (C11, C12, C16, C18, choke). You could see that the spike still existed, the sensor showed drifting as well. Any suggestions?

Hey Specia,

Can you clarify that this is a scope capture and not a spectrometer capture?  Also can you comment on where you are probing the signal?

Hi Patrick,

This is the scope capture. 

This is the way i probe the signal.

You could see I also use two cables with shielding, it is 25 cm length, when i unplugged the cables it's normal, but when i plugged the cables i got spike issue.

Then i tried to add capacitors filter and choke, like in the eval-module schematic, but yah it's getting worse. Any suggestions?

I put electromechanical relay, because i have a plan to switch the cable connection for another sensor. I guessed this issue is related to the switching as well. Do you have suggestion a switching without external mux or relay? Such can use the internal mux of the sensor. But even it has internal mux, when it will be connected to an op-amp to generate the active shield signal, it will need external mux as well right?

The spike looks like weird since it was negative value only, why the spike is not positive value?? Are you noticing something?

Hello Specia, 

Patrick is out today but will be back Monday to review your additional information. 

Thank you for your patience. 

In the meantime, I can comment on the following: 

specia special said:

Do you have suggestion a switching without external mux or relay? Such can use the internal mux of the sensor. But even it has internal mux, when it will be connected to an op-amp to generate the active shield signal, it will need external mux as well right?

The FDC2214 internally switches between the 4 channels of the device. If you were to use all 4 internal channels, you would need one op amp per channel. Having an external mux is acceptable for this device as long as your mux has a high enough bandwidth to allow the sensor signal to pass through properly. 

Best Regards, 

Justin

Hey Specia,

Thanks for clarifying your setup.  Perhaps you are suffering from contact bounce with the relay.  Im thinking that depending on how cleanly the relay closes, it might distort the signal in such a way that the signal level thresholds are not met during the bounce period, thereby reducing the perceived frequency of the sensor.  As the reference signal is clean and consistent, the fsensor/fref frequency will decrease.  You might see if you can have your rigol setup to catch runt pulses or trigger per pulse width to catch the debounce event.  For fixing the issue, you might try the circuit here. Or you could just keep track of when you switch the relay and delay before reading result and stop reading before switching.

Hi Justin,

Thanks for your response. I've been just back from our company long holidays.

Hi Patrick,

I've been just back from our company long holidays. Anyway...the circuit you suggested is for MOSFET relay while i am using electromechanical relay like in the setup above. And I am connecting the sensor to normally closed contact FYI (pardon i forgot to give a note on that). Any suggestions?

btw I've also ever tried a setup like this :

1. used electromechanical relay, used shorter cables around 3 cm, and didn't get the spike issue anymore
2. didn't use electromechanical relay, i connected longer cables to the FDC2214 directly around 25 cm, and didn't get the spikes issue as well

3. used electromechanical relay and longer cables around 25 cm, yah the issue was coming up

Is this issue related to the long cables? I've also tried to add the filter capacitors like in the eval-module schematic, but yah it didn't fix the issue still got spike issue as shown below for the reading data.

Hello Specia,

I think the issue is related to your relay.  Are you switching the relay at all between measurements?  If so, how long do you wait to collect measurements after the relay switches?

Hi Patrick,

No, I'm not switching the relay at all between measurements, FYI I connected the sensor to normally closed pin of the relay and none of switching

I found something here. It's LDC family, but the LC-tank circuit is pretty similar. Can I use the equation 1 and 2 (page 2) for FDC2214?

Anw still about spikes issue, I've tried to add capacitor in series with the relay (without switching exactly, use the normally close pins), the schematic is shown below.

and I got signal like here below.

you could see the amplitude of the signal is lower than amplitude of the signal picture i posted previously. After I add capacitor in series in the relay input pins, the amplitude became 1.6 V w/ 1MHz (while in the signal picture i posted previously, the amplitude is 3.84 V w/ 570kHz).

and this is the reading data when the amplitude became 1.6V

And didn't find spikes issue.

doesn't it show that the spikes issue is coming from out of ideal range (the ideal range = 1.2 - 1.8 V)??

Hello Specia,

I need to check with my team on whether you can use those equations for this device.  Staying within the range specified in the Electrical Characteristics table definitely seems like something you should strive for.

Hey Specia,

We think that those equations should apply here.  FDC2214 leverages part of the same architecture found in the LDC parts and we think this should be similar.  Original designers have long since left, so it is a little difficult to definitively prove.

Hi Patrick,

Alright, I thought so for those equations.

Anyways....still about the spikes issue, could you tell me how to check that this issue is not coming from our code?

Hey specia,

So my understanding is that you have done a few combinations:

1. electromechanical relay + short cables : no spikes

2. long cables : no spikes

3. electromechanical relay + long cables : spikes

4. high pass filter caps + electromechanical relay + long cables: no spikes

Im thinking if it was a matter of post processing error, you should see the spike issue for each test case.  Do you have a spectrum analyzer you can hook up in front of the relay?  Perhaps some external noise is coupling in sporadically.

Hi Patrick,

for number 1 - 3, that's right
I will correct for number 4. high pass filter caps + electromechanical relay + long cables : SPIKES

Alright, currently I am still preparing the spectrum analyzer.

Btw, here I've also tried to read the sensor data used the eval-module MSP430F5528IRGC. Even connected the electromechanical relay + long cables, surprisingly I didn't get the spikes issue.

The different condition is when I found the spikes issue, I used my personal board w/ STM32F4, and I connected the SD and ADDR pin of FDC2214 to GND. While for the eval-module MSP430F5528IRGC, the SD pin is connected to P1.2/TA0.1 and the ADDR pin of FDC2214 is connected to P1.3/TA0.2.

Are you noticing something with my connection?

I also have another question.

Datasheet says for current drive setup, it has reserved bits 000 0000 0000.

I tried to set the current of  channel 0 to 0.146 mA (0x7800). It means the value of register address 0x1E should be 0x7800 (0b0111 1000 0000 0000). Then i set it to 0x7800. But when i tried to read the register, it became 0x7bc0 (0b0111 1011 1100 0000), it means the reserved bits are 011 1100 0000. Is it okay??

Anyways, what is the meaning of reserved bits in the datasheet? How if the value doesn't match the datasheet?

Hello Specia,

SD and ADDR pin connected to gnd should not be an issue.  I would expect that to only introduce errors if you have a weak ground connection.  Do you use a ground plane on your board?  If so, do you route immediately to that ground plane, or do you have a long thin ground trace going to the ground plane?  Do you have decoupling caps on the supply near the FDC2214 and the STM32F4?

Hi Patrick,

Why do you guess that this spikes issue might be coming from the weak ground connection?

Yass, I do have the decoupling caps near those items exactly