• State
  • Locked Locked
  • Replies 11 replies
  • Subscribers 46 subscribers
  • Views 2158 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.

DC Drift and Low Frequency Noises during Static Photodiode Measurements with AFE4403 EVM

Brian Leung
Intellectual 325 points
Other Parts Discussed in Thread: AFE4403, AFE4403EVM

Dear Sir,

I am trying to use AFE4403 EVM for some static photoelectric measurements, checking light intensities through solvents with different concentrations. Results seem ok but I noticed that there are DC drifts and Low frequency noises (0.x to 2 Hz) on the ADC output voltages. The DC drift was small (around 1-2 mV in 30 secs with Rf = 1M ohm, Second Stage Gain of 4 ) but would affect accuracy of my measurements significantly. I just wonder if possible to reduce/remove these dc drifts and low frequency noises. 

Regards,

Brian

  • 0
    TI__Genius 12575 points
    Hi Brian,
    When you say ADC output, are you referring to LED data or LED-Ambient data?
    If your ambient light is changing slowly, you will see a drift in LED data. So to cancel the drift due to ambient you can always use LED-Ambient data.
    For e.g.
    1) Use LED2-ALED2VAL ( Register 0x2E) data, if you are using LED2 or,
    2) Use LED1-ALED1VAL ( Register 0x2F) data, if you are using LED1

    Regards,
    Prabin
  • 0 in reply to Prabin Yadav
    Intellectual 325 points

    Hi Prabin,

    Actually I am using LED minus Ambient already. But it's either increasing or decreasing slowly versus time ..say a few mVs in 30 seconds , for example. I am not sure why the baseline is either going up or down slowly but my setup is basically static . 

    Regards,

    Brian

  • 0 in reply to Brian Leung
    TI__Genius 12575 points

    Hi Brain,

    Following are the points that we could think of as the reason for the DC drift:

    1) Non linear behavior of LED: If the duty cycle of LED is higher, the temperature of LED might increase. This might change the intensity of light for a given LED current and finally cause the ADC output to drift. You could try changing the duty cycle of LED and see how it effects.

    2) Just to confirm its not ambient that is changing slowly, you can turn off LED and observe AMB data as well as LED-AMB data (i.e. AMB-AMB in this case). Drift in this case will point us towards changing ambient as cause.

    3) Movement in solvent: If the solvent under test moves a bit, the light received by the Photodiode will change resulting in DC drift. This is something related to mechanics of your setup. To rule out this possibility you can use a reflective surface and redo the measurements.

    Could you please share some information such as Duty cycle of LED, PRF(sampling frequency)?
    Regards,
    Prabin

  • 0 in reply to Prabin Yadav
    Intellectual 325 points
    Hi Prabin,

    Thanks for your advices.

    I agree that the LED temperature may rise but I did a test yesterday that in each cycle, I switch on the LEDs for ten seconds and then switch them off for around 100 seconds ...and repeating this cycle for 30 times to see if the output voltage still rises or not..The result was..it still kept rising for LED1 but decreasing for LED2.

    Currently, the sampling frequency is 200Hz but I need to check the duty cycle . I am driving the LEDs with around 9-10mA.

    And I will try your suggestion (2) early next week. As for movement of solvent, I think it's static and wont be generating monotonic increasing/decreasing photo detector output though.

    Regards,

    Brian
  • 0 in reply to Brian Leung
    TI__Genius 12575 points
    Hi Brian,
    Yes, you are correct regarding the movement of solvent, even though I have following quick questions.
    1) When LED is switched on for 10 sec, does the ADC samples during this 10 sec window drift? Or the average of 10 sec data over time drifts?
    2) Is LED1 always rising and LED2 decreasing?

    Regards,
    Prabin
  • 0 in reply to Prabin Yadav
    Intellectual 325 points

    Hi Prabin,

    1) yes, it's either increasing or decreasing within that 10 secs and the same if I took average for each cycle. 

    2) yes, LED1 is always increasing , while decreasing for LED2

    3) the led on time is 1/10 of the full cycle (which is 6999 for the register@0x0a and 5000 for register@0x09 respectively)

    Regards,

    Brian

  • 0 in reply to Brian Leung
    TI__Genius 12575 points
    Hi Brian,

    Based on your reply can we try following tests:
    1) Try with lower duty cycle for both the LEDs.
    2) Lets observe LED, AMB and LED-AMB data with both the LEDs switched off.
    3)Results with the LED1 switching timer swapped with that of LED2. Does "LED1 is always increasing , while decreasing for LED2" still holds true?

    Regards,
    Prabin
  • 0 in reply to Prabin Yadav
    Intellectual 325 points

    Hi Prabin,

    the following is captured from AFE4403EVM GUI ...with one percent Duty Cycle (RF = 100k, Cf=5pf, same TIA for both LED1 and 2, no stage amplifier enabled)  while both LED1 and 2 were off.

    Obvious that the Ambient output was drifting ..actually any drift ideas of AFE4403? how many ppms/degC? and why ambient output drifts? 

    Regards,

    Brian

  • 0 in reply to Brian Leung
    TI__Genius 12575 points
    Hi Brian,
    This drift in Ambient can be because of surrounding lights and is not related to AFE4403. To verify this you could do following tests:
    1) Enclose the sensor (LED + Photodiode ) in black box and repeat the same experiment (i.e. with LED switched off)
    2) Physically disconnect the photodiode and repeat the same experiment.

    By the way, how is the result with lower duty cycle with LED switched ON?

    Regards,
    Prabin
  • 0 in reply to Prabin Yadav
    Intellectual 325 points

    Hi Prabin,

    1) less drifts if further covering the light sensors from ambient light

    2) yes, basically no drift if disconnected the DB9 port of the EVM

    3) Actually the screencapture i posted yesterday was with 1% duty cycle already. 

    I speculated that there are capacitance somewhere that being charged by dark current due to either ambient/LEDs. 

    The one from ambient is creating the downward drift and positive drift from the LEDs. 

    Regards,

    Brian

  • 0 in reply to Brian Leung
    TI__Genius 12575 points
    Hi Brian,
    I thought, the screenshot that you sent yesterday was with LED switched OFF. We can check with LED ON and with lower duty cycle.
    Also, how is sampling timer for LED and Ambient(AMB) phases placed? For better cancelation of AMB, you should place the sampling timers for LED and respective AMB adjacent to each other.

    Regards,
    Prabin