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.

AFE5818: TCG slope non-linearities

Part Number: AFE5818
Other Parts Discussed in Thread: AFE5808A

We have been trying out the AFE5818 using the eval board and we have run into an issue. When we apply a slope to VCNTL, we see that there are non-linearities at lower gains. The steeper the slope, the worst the error is. Here is what we did:

  1. We changed C201 to C202 on the Eval Board to 100pF.
  2. We statically measured the voltage required on VCNTLP to get a specific gain in 1 dB steps in order to build a VCNTL vs gain curve with a 38 dB range. We drive the AFE5818 gain control in signal ended mode.
  3. We use this curve and apply it using a function generator at a specific slope (for example 5dB/us)
  4. We applied a signal at the input and analyzed how it increases as the gain increases.
  5. The results for a 5dB/us slope is shown below.

The first image show our VCTRL curve and the sine wave input. The second shows the rectified sine wave converted to a logarithmic scale with the ideal gain curve in red. The gray dashed gain curve is taking into consideration that there is a 0.35us delay. The gap between the gray line and the sine wave is our concern. Between the start and about 6us, there is a significant difference between the two. The slope at the start is very slow. We were expecting a delay, but not such non-linearities.  The 800kHz bandwidth for the VCTNL inputs should be sufficient for the slopes we are using. The third curve shows the same slope, but with the gain but with starting at a gain of 10dB instead of 0dB. This curve is ok. Starting from 0dB gain seems to be the problem.

Is there something we can do about this?  Are we doing something wrong?    

Thanks,

  • Hi Eric,

    Welcome to TI E2E forum.

    Regarding your question regarding the TGC non-linearity of AFE5818, here is the response from our Systems Engineer.

    The low gain non-linearity is from the below region in the datasheet. The curve is more linear at higher gains.

    In order to calibrate it, you may need to adjust the DAC programming.

    Since AFE5808A and AFE5818 share the same VCA die, you should be able to use the below table to calibrate.

  • This has already been compensated for. You can see that the VCTNL voltage that is applied is not a straight line. Moreover, we stay out of the the very non-linear regions past 1.2V   

    If a slow slope (less than 1dB/us) is applied, there is no problem. It is when a steeper slope is applied that the problem occurs.

  • Here is a slope at 0.5dB/us. You can see that there is no problem.

  • Hi Eric,

    Thanks for sharing the data with the lower slope.

    As we are looking into this, would it be possible to take the data with the 100pF caps removed?
    Also can you please share the calibration curve for the VCTNL vs gain curve that you statically measured to feed to the FG?
  • Removing the 100pF capacitors had not effect. Here is the table we used.

    Time (us) VCNTL (V) Gain (dB)
    0 1.184 0
    4.5 1.184 0
    4.7 1.123 1
    4.9 1.073 2
    5.1 1.042 3
    5.3 1.012 4
    5.5 0.987 5
    5.7 0.954 6
    5.9 0.915 7
    6.1 0.879 8
    6.3 0.85 9
    6.5 0.824 10
    6.7 0.792 11
    6.9 0.754 12
    7.1 0.719 13
    7.3 0.69 14
    7.5 0.665 15
    7.7 0.641 16
    7.9 0.609 17
    8.1 0.576 18
    8.3 0.547 19
    8.5 0.521 20
    8.7 0.498 21
    8.9 0.475 22
    9.1 0.445 23
    9.3 0.413 24
    9.5 0.385 25
    9.7 0.36 26
    9.9 0.337 27
    10.1 0.313 28
    10.3 0.285 29
    10.5 0.255 30
    10.7 0.227 31
    10.9 0.201 32
    11.1 0.175 33
    11.3 0.148 34
    11.5 0.114 35
    11.7 0.078 36
    11.9 0.043 37
    12.1 0 38
    20 0 38
  • Hi Eric,

    I am closing the thread since we took the conversation offline and addressed the issue.