Huge offset voltage with ADA4897-1

I have a standard inverting amplifier stage with Rf = 100k and Rg = 10k (A = -10 V/V).  The amp is powered from +/-5V and the positive input terminal is connected to 0V.  Pins 1 and 5 are open.  Pin 8 (nDISABLE) is also left open.  Very straightforward setup.

The problem is that the output has a DC offset of -2.2V, but otherwise behaves as expected for AC signals.  Tying the disable pin high has no effect.  This was repeated on three ADA4897-1 parts and all were identical.  Inserting a bias compensation resistor of 9kOhm between positive input terminal and ground brings the DC offset to about 100mV.

Without the bias compensation resistor, I swapped out the part for an AD8610 and the offset disappears and the circuit behaves normally.  Interestingly, the offset voltage is absent when I use the ADA4897-1 in a basic non-inverting configuration (Rf = 247, Rg = 13, A = 20 V/V).  Did I just receive a batch of bad parts, or is there something I'm missing specific to these components?

Thanks,

Jason

  • 0
    •  Super User 
    on Aug 8, 2013 7:07 AM over 7 years ago

         Hello Jason,

         The relatively high resistances of the inverting configuration (Rf = 100K) along with stray capacitance is likely allowing enough phase lag at 250 MHz to make the circuit oscillate.  Unless you have a really fast oscilloscope you may not even see the oscillation, but the unusual offset is a clear symptom.  This also explains why both  the non-inverting circuit (with much lower resistor values) and the AD8610 (with 10x lower BW) do work.

         Best regards,

         Bob

  • Thanks for responding so quickly.

    Bob, I don't see any signs of instability up to 100MHz, but I can inspect the circuit with a higher frequency scope.  That's a good suggestion.  I initially had a feedback capacitor in place to limit the bandwidth.  This amp is fully compensated (stable gain of 1), so I didn't consider that to be a potential issue.

    Harry, I do understand the noise impact on these component values.  The values in the inverting amplifier configuration is not intended to be anything but a sanity check.  If I can't get it to work in this configuration that would likely mean something else is going wrong (bad layout, bad part, power supply, etc.).

    I will test the inverting amp configuration with lower values (Rg = 10, Rf = 100) and report back here.

    -Jason

  • Harry, I can also add what I'm trying to work toward.  The final configuration should be a gain of 10-20x up to 400 kHz.  The first stage will likely have to be ~10x due to GBP limitations.  The noise limits will be the source impedance (in the range of 1k-10kOhm real, 100-300pF imag).  I'm considering multiple configurations, including a charge amp.  Noise current density is the one of the most important noise contributors at these source impedances.

  • Jason,

    Could it possibly be the input bias current of typ. 11uA, flowing in the 10k resistor, making 110mV then times 10 gain? As you observe, using a current comp resistor in the other lead reduces the offset a lot. Just like the uA709 op-amps of long ago. They used to oscillate at high frequencies too.

    The other amp you mention has pA of bias current - a much different beast.

    Simon

  • Simon,

    The current bias on these amplifiers was indeed causing the large voltage offset at the output.  My source impedance is too high to use with the ADA4897; however, some of the amplifiers suggested by harryh proved to be better suited anyway due to the lower current noise spectral density (<1pA/rtHz at high frequencies).

    Thanks for all of your comments,

    Jason