Post Go back to editing

AD8422 oscillation at output

Hi!

I use the AD8422 In-amp in two different places in my board. Both AD8422 uses dual supply of about -12V and 12V. Both are connected to an analog switch and then to the input

of a LT1468 that is used as a buffer to drive an ADC.

One is used to measure a voltage within -5V to 5V and the other one to measure a current -50mA to 50mA by sensing the voltage over a 100R resistor.

Both get oscillations at around 750kHz at some voltage/current levels. For example the voltage output is good between 0-5V but oscillations occur around -2V, but not for the complete range -5V to 0V.

The current measurement was oscillating all the time if I remember correctly.

At the output I used a 50R resistor and then 100pF to GND.

Increasing the 50R to 10k oscillations stopped, using 1k still no oscillations. With 500R it start oscillate again.

Finally I used 1k and 1nF to get a suitable low-pass filter for my application and in this case no oscillations either.

The question is why I see these oscillations from time to time and most importantly how I can make sure I not get them.

For the current measurement case I also tried to change the gain from 1 to 2 before changing the low-pass filter on the output, but no change in oscillation freq or amplitude.

I got the same behavior on two different board with the same solution (increase resistance in output low-pass filter).

Best regards

Joakim

  • Hi Joakim,

    Can you share the schematic file of your circuit? We'll share this to our apps engr and get back to you. All the best!

  • Hi!

    Sorry for late response.

    I must check with customer what I am allowed to share. Is it possible to share it to your team privately?

    What can be added is that after each output the signals go to an analog switch ADG5233, after that

    the signal can go to either an ADC driving circuit based on your application note dn1032fa.

    The other possible way is to the negative input of an OP, LT1007. A 3.4k resistor is used in series with the

    OP input and the analog switch output.

    Best regards

    Joakim

  • Hi Joakim,

    Thank you for reaching out to us.
    While waiting for the schematic, I like to clarify some information about the customer's circuit in order for us to be able to help efficiently.

    May we know what's the target bandwidth of the customer's system? specifically going in to the AD8422? I liked to ask since it was mentioned that the suitable low pass filter (1k and 1n) is way below the frequency you're originally looking at (~750kHz).

    I'm also interested about the input signal's nature (Frequency? ), is it coming from an equipment going through an analog switch or sensors? For the one configured for current sensing, what's the current range the customer is trying to measure?

    Can they also provide scopeshots of the input and output on the AD8422 when it's oscillating and when it's not oscillating? It will be better if they can probe it as close to the pins as much as possible. Checking the AD8422 datasheet's small signal response, having a cap load of 100pF is introducing a lot of ringing on the response, additional parasitics may cause this to increase further and may make the part oscillate. Given this, can we confirm if the customer is using a PCB board when performing the test? or just a vector board?

    You can include the information with the email you're going to send.

    Thanks Joakim.

    Best Regards,
    Dann

  • Hi!

    The target bandwidth is about 100kHz, I used higher bandwidth in the beginning to see how different parts contribute to the performance and not "hide" unexpected behaviors as high frequency oscillations and similar.

    The input signal is coming from remote voltage measurement, but in this case the remote point was very close to the PCB. The current is always measured directly on the board and typically is between -50mA to 50mA, or -5V to 5V at the AD8422 input terminals.

    The remote sense is also between -5V to 5V.

    These signals are passed analog MUX to either measure it with and ADC, or to the input of an OP circuit that regulates the output voltage/current to a given set point.

    First I was expecting it was the feedback from the OP that somehow started to oscillate but I was not able to measure any oscillations on the voltage output signal or on the input pins of the AD8422.

    I saved a screenshot from the oscilloscope (blue trace) directly from the output pin of the AD8422 before I changed to 1k and 1nF on the output filter. Also not shown here, when it works correctly the output signal was clean. Note also that the expected (probably) DC signal is present as an DC offset in the oscillation.

    The PCB is an 8-layer board.

    Is there some limits on the allowed min/max C-load on the output to keep the device stable that may introduce the oscillations?

    Best regards

    Joakim

  • Hi!

    I had another board without the changes so I did some measurement before updating the output low-pass filter. With R414=50R and C409=100pF at output of AD8422.

    I connected a signal generator between REMOTE_SENSE_P and REMOTE_SENSE_N=ISO_GND.

    Output from AD8422 is measured on a VIA before the signals goes to the LP-filter, green trace below. The brown trace is the signal generator output at the remote sense input terminals.

    (Due to non-ideal grounding and an isolated DC/DC there will be some HF noise pickup that can be ignored)

    With a triangle wave I see the oscillations occurs around -1.5V to 0V.

    AC ripple when input is fixed at 2V:

    AC ripple when input is fixed at -2V:

    The response to a step input looks nice without any ringing in the non-oscillation range:

    The response to a step input have oscillations but no extra ringing in the oscillation range:

    Change components to: R414=1kR and C409=1000pF at output of AD8422, the output does not have oscillations anymore.

    Not sure if the above can help? I was expecting some ringing at the oscillation frequency for the

    step input but it was no obvious ringing in the measurement.

    Best regards

    Joakim

  • Hi Joakim,

    Thank you very much for this screen captures.
    I also find it odd that it only appears on specific voltage levels. 

    I'll investigate this further and let you know what I found soon as I can.

    Thanks and Best Regards,
    Dann

  • Hi!

    Thanks for looking into this.

    In the meantime I did some measurements of radiated energy from the PCB.

    with onboard isolated DC/DC

    Without the isolated DC/DC.

    In both cases there are 3 other DC/DC working and a couple of LDOs for generating the different supply voltages.

    We have some peaks close to the observed oscillation frequencies but not spot on. Just trying to figure out how the oscillations starts and how to get ride of them.

    Best regards

    Joakim