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.

INA849: Overshoot control

Part Number: INA849
Other Parts Discussed in Thread: ADC3643EVM, , INAEVM-ALT-SO8, ADC3643

Hello TI Team,

I am currently working on the INA849 that I would like to use to feed the input of an ADC3643EVM board.

It is connected to a square wave generator with typical amplitude below 50 mV and a 1us with very sharp edges. I used the regular application schematic of the INAEVM-ALT-SO8 board : 

I didn't show the input section as it is part of a design subjected to confidentiality, but lets just say that the differential input voltage is very low in amplitude, so that the instrumentation amplifier output is not saturating.

Despite the output low pass filter, I get very large overshooting spikes lower than 100 ns wide and higher than 100% in voltage amplitude, that would be damaging to the ADC input. I tried lowering the filter cutoff frequency but it resulted in an increase of the overshoot recovery time, while the overshoot rising remained sensibly the same.

Do you have any suggestion to remove, or at least very significantly tame the overshoot, so that the input of the ADC remains safe ? Also, what would be the best approach to protect the ADC input, would a simple Shottky diode be sufficient ?

Thanks,

Best regards.

  • Please disregard the +/-2.5V power supply values, in real application it is +/- 5V. I need to correct the schematic.

  • Hi Quentin,

    can you show scope plots of the input signals of INA849 and the output signal?

    Eventually, a weak RC input low pass filter can help to make the edges of square wave less steep and by this decrease the overshot.

    Kai

  • Hi Kai,

    Thanks for your answer. I cannot disclose too much details here, but we are already tried adding slopes and we still get the large overshoot as shown in the picture below :

     

    I am considering using a clipping opamp circuit like this one : 

    The overshoot in itself is not an issue for the application, but I want to avoid frying the input of the ADC to which the signal is fed.

    As for the input low-pass filter, it is not an acceptable solution since we need to avoid loading the source as much as possible, hence why a FET input instrumentation amplifier was selected.

  • Hi Quentin,

    I am unable to duplicate the overshoot in INA849 with Gain of 10V/V (20kHz square wave). 


    INA849 Overshoot 09022021.TSC

    Could you check your probe compensation in your scope? I want to rule out the possible artificial effect. 

    Best,

    Raymond

  • Hi Raymond, all the connections are done through SMA cables directly to the scope, there are no probes and no probe compensation engaged.

  • Please note that one of the input is connected to a 50 Ohms source and the other to an high impedance load (> 1 kOhms). A 50 Ohms resistance is connected between the two inputs as well.

  • Hi Quentin,

    this is normal small signal step response, as shown in the datasheet.

    Kai

  • Hi Quentin,

    I'm a bit confused. Please tell us why you have chosen an INA for this purpose.

    With what signal do you want to feed the ADC3643? With a square wave? With varying amplitude? And why must the square wave have steep edges? What parameter of square wave do you want to analyze with the ADC? Would a peak detector and a much slower ADC also do?

    Kai

  • Hi Kai,

    Thank you for your answers. The small signal responses overshoot in the datasheet are definitely not as dramatic as what I have observed, but it may therefore be better to increase the gain.

    I use the ADC as a voltmeter to measure the amplitude of the square wave when it is stabilized. I needed a fast instrumentation amplifier to measure precisely these steps amplitude without influencing the source and the load. Otherwise I would have used a differential amplifier but these have 50 Ohms inputs which is not acceptable in my case.

    Going slower is not an option, we need to be as fast as possibly allowed by the INA to get a good, flat signal section to measure with the ADC.

    Steep edges are not mandatory at all and we actually already implemented smoother one that improved the response, but overshoot is still there. I will consider increasing the gain and decreasing the input amplitude, and introduce shottky diodes at the output to protect the ADC.


    Unfortunately I cannot disclose too much informations in this public forum... But thank you for your help.