• State TI Thinks Resolved
  • Locked Locked
  • Replies 6 replies
  • Answers 2 answers
  • Subscribers 30 subscribers
  • Views 556 views
  • Users 0 members are here
Support feedback
Options
Options
Related

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.

LDC1614: Instability with Multiple LDC161X and internal frequency reference

Samuele
Prodigy 40 points
Part Number: LDC1614
Other Parts Discussed in Thread: LDC1612,

Hello,

I am using two LDC1614s and one LDC1612 to detect the presence and positioning of a metal target.
The system sequentially enables one sensor at a time after putting the other two in standby (SLEEP_MODE_EN = 1).
An external crystal (40MHz) supplies the same clock to the three sensors.
The 10 LC sensors have the same layout, the same capacitor and are driven with the same current (AUTO_AMP_DIS = 1, RP_OVERRIDE_EN = 1 and DRIVE_CURRENT = 17). The measured frequency is very similar (~ 3MHz) as well as the conversion result (2037697 <DATAn <2049399 +/- 3 with OFFSETn = 0).
Now, if I select internal clock (REF_CLK_SRC = 0) the absolute values ​​decrease due to the theoretical increase of the reference frequency (43.4MHz typ VS 40MHz). The values ​​are similar for the two LDC1614 which go from just over 2 million to 1.8 million with the two DATA of LDC1612 which drops to just over 1.7 million. This, probably due to a higher value of the internal oscillator.
This not be a problem for the functioning of the sensor.
The problem is that the instability greatly increases with DATAn values ​​that vary even by +/- 100.
Such unstable values of DATAn ​​at rest and very different when approaching the target (same distance) are not acceptable for the functioning of the sensor.

Do you have an explanation or a suggestion that can solve?
I am available to provide further details.

thanks

Best regards,

Samuele

  • 0
    TI__Expert 7599 points

    Hello Samuele,

    Can you provide more information on the instability you are describing ? 

    If I understand the system correctly you have 3 Sensors and the 3 sensors are driven by an external clock ? The External clock is provided with some sort of a mux and controlled and managed between active and non active channels ? 

    The Internal clocks specs are as listed below, the clock can vary from 35 to 55 Mhz and can drift. 

    For a stable high precision measurement using LDC devices the recommendation is to use a dedicated external clock for each of the sensors. The output of the LDC device is a function of clock frequency for LDC1614. Switching between int and ext clock may cause instability. 

  • 0 in reply to Arjun_Prakash
    Prodigy 40 points

    Hello Arjun Prakash,

    with the external clock the values ​​of the 10 DATAn vary, at rest, only by a few lsbs. For a maximum variation of +/- 3. Placing the target above the various sensors, at the same distance, the maximum values ​​are comparable and have small differences.

    With the internal clock the values ​​of the 10 DATAn vary, at rest, by several lsbs. For a variation greater than +/- 100. By placing the target above the various sensors, at the same distance, the maximum values ​​are not comparable and have important differences.

    The external clock is connected to the CLKIN pin of the three sensors. It is never turned off. There is always only one sensor active while the other two are placed in standby (via SLEEP_MODE
    _EN in CONFIG). What is multiplexed is the SDA signal.

    Differences in the values ​​of the three different sensors were expected. This is due to the diversity of the reference frequency (internal clock). The values ​​obtained are consistent. DATAn = ~ 20M with ~ 3MHz of sensor oscillation (LC) and 40MHz of external clock which becomes ~ 18M with internal 43MHz or 17M with 45MHz for the two channels of the LDC1612.
    Just as a "slow" drift has been contemplated, due to variations in temperature or voltage, which would not be a problem for the algorithm.

  • 0 in reply to Samuele
    Prodigy 40 points

    Hello Arjun,

    I'm available for further details

    Thanks

    Samuele

  • 0 in reply to Samuele
    TI__Expert 7599 points

    Hi Samuele,

    Thank you for the update.

    As you have pointed out that variation with external clock is stable with less variation I do recommend that all your sensors be driven with external clock. 

    As I have pointed out in my previous message the output data frequency is a function of Fref which is internal or external. A high precise external crustal oscillator with tighter specs than internal clock is an option and that has been made available for such high precision applications. 

    One other options to control the variation even after an external clock is provided is to calibrate and use an unused coil as a sensor. This can compensate for any thermal variations in the system and may help compensate for thermal expansions of the PCB. There is also a expansion of the target that needs to be considered. As ambient temperature effects the thermal properties, thermal expansion of the material also needs to be considered here. 

    Let me know if this helps!

  • 0 in reply to Arjun_Prakash
    Prodigy 40 points

    Hi Arjun and thanks for the reply,

    however the variations to which I refer are very rapid variations not due to the variation of the temperature. Let's talk about important differences between two consecutive readings. This regardless of the target presence.
    I don't know if they may be due to a power supply noise. Power supply seems is stable. Three capacitances (1uF, 100nF and 10nF) are close to the power pins of each LDC161X sensor.
    Is internal oscillator more susceptible to voltage variations than external quartz?

    Regarding the settings of the registers, these, with the exception of the REF_CLK_SRC bit in the CONFIG register, are exactly the same.
    Analisyng with the oscilloscope the stabilization time of the oscillation does not vary with internal or external reference. It shouldn't be related to the internal oscillator. In any case I also tryed to increase the value of SETTLECOUNTn but without results.

    The sensor has already been built and tested with good results but there are problems with irradiation (EMI). The first solution thought was to use the internal oscillator by removing the external one.
    The sensor is resin coated and at the moment, for these tests, the external oscillator is still connected and functioning. Could this be a problem?

    Do you have any suggestions or should I discard the use of the internal oscillator?

    I'm available for further details

    Thanks

    Samuele

  • 0 in reply to Samuele
    TI__Expert 7599 points
    There could be several factors that impacts the accuracy. Noise in the system can certainly be one of them. Noise on the power, Gnd can be one of the factors. For highest accuracy in your system you are able to get it by using external clock and I would recommend this as the Clk parameters are typical and it’s possible that noise in the system can be coupling into the internal clk. In regards to EMI, we do have an app note that provides guidance on recommendations for EMI. www.ti.com/.../snoa962.pdf . I can’t comment on resin coated as I am not sure about it. In terms of accuracy as you are getting good results with external clock I would recommend to use for external clock and consider adding chokes or ferrrite beads to address EMI problems.