AD8338 AGC Discontinuous GAIN


In AGC mode, I apply a slowly-AM modulated Input signal (CH1 below).

As the input amplitude increases, the output (SENSOR_OUT = CH2 below), reaches the VOUT_RMS target value of ~250mV, the AGC (attenuation) engages suddenly accompanied by a large spike in SENSOR_OUT level.

As the input amplitude falls, the SENSOR_OUT fails to track the Input (time  = 0-1.8sec below).  In this interval the observed SNR falls rapidly.  That is, the AD8338 output contains in-band noise amplified to 250mVrms.

The Schematic:

I welcome any suggestions as to what causes this behavior and how to correct.


  • 0
    •  Analog Employees 
    on Aug 19, 2019 8:00 AM over 1 year ago

    Hi PeterM,

    I haven't noticed anything wrong with the circuit.
    Though, may I ask the details regarding the AM input signal,
    What's the modulation and carrier frequency?
    Are you using the AD8338 evaluation board on your setup?

    I'll still check if I missed something on the circuit.
    Apologies for the delayed response.

    Best Regards,

  • Hi Dann,

    The AM input carrier  is 17kHz.  The AM is a 0.1Hz saw tooth modulating 99.9% to create this slowly-varying test signal. 

    The input signal originates at a single-ended signal generator.  I pass this signal through the AD8338 Evaluation Board (configured for unity gain and single-ended to differential conversion).  Each output (true, complement) of the Evaluation board contains in-line 40dB attenuation (80dB total) before being fed into the circuit above.

    Additionally, at R100 / R103 divider, I reduced the values to 1K / 5K under the suspicion that the high impedance at this node was coupling noise in.  -- This change showed no improvement.

    At low input levels, the circuit above provides the expected 80dB gain (GAIN = 100mV).  When AGC engages, GAIN jumps by ~500mV corresponding to about 40dB.

    Thanks for looking into this.

    Best regards,


  • 0
    •  Analog Employees 
    on Aug 29, 2019 3:31 AM over 1 year ago in reply to PeterM

    Hi PeteM,

    Thanks for the information.
    It seems that the signal frequency is slow and is accommodated by the bandwidth set by the feedback capacitor at the output of AD8338.

    Have you tried removing the probes from TP102?
    The probe capacitance may be affecting the AGC response time through the DETO pin.
    Can you also look and have a scope shot of the signal at the output of ADA4940-1 which is used as the single-ended to differential ended amplifier on the AD8338's evaluation board?

    Thanks PeteM.

    Best Regards,

  • Hi Dann,

    Sorry for the slow reply, I was diverted to another task and lost some of my test setup to another program.

    Yes, I tried removing the TP102 probes.  The problem persists:  It existed before I probed that point, and the probe's ~10pF load is negligible compared to C110.

    Here's the scope shot of the Input (CH1), SENSOR_OUT (Ch2), and AD4940 output (CH3-Diff):

    I don't see anything anomalous in the AD4940's output. 

    Thanks for your help so far.  Please let me know if you see anything else.