Post Go back to editing

VREF rise time

I'm working with an AD7766 design in which I just discovered the VREF is being supplied with a reference IC and amplifier circuit with a pretty big time constant (~1 second).  Therefore, I'm seeing the VREF take about 6 seconds to 'fully settle' on power up.

In contrast, the AVDD settles pretty quickly (~500uS) after power is applied.

I've attached a graph illustrating what I'm describing.

My question is, am I risking any long term damage to the chip leaving the design this way? The board that this ADC is being used on is fairly mature and has a lot of usage as we're getting ready to move to production with it. I'd rather not change any characteristic of the VREF and it's filtering if I don't have to, due to the amount of testing, but will if I'm risking damage to the part.

After VREF has settled, will the subsequent reset (SYNC/PD line goes low, then high), address any oddities in the power supplies coming up at different rates?

Scope plot legend:

Orange = AVDD

Blue = VREF

Black = VIN+

Green = VIN-

Thank you.

  • Hi agporsch,

    The vref input is typically a much slower input to power-up and settle, due to the larger capacitors on this node, so it is not unusual to see the longer rise time of this supply/input compared to the Avdd and Dvdd supplies. As long as you do not exceed the absolute maximum ratings as specified on page 8 of the AD7766 datasheet then no damage should occur. 

    However, I do see in the plot that the Vref and Vin+/- input are recorded as starting at -7V and -8V. If this is relative to the AGND or REFGND in the system then this would be a concern. The maximum negative voltage allowed on these pins is -0.3V relative to AGND.



  • Thank you for the response. While I agree, due to the suggested placing of a 100uF cap on the VREF line, generally speaking, I would expected to see a 'larger' time constant.

    However, assuming a low impedance of the amp driving the VREF (say ~50 ohms), I would expect a time constant on the order of 5 milliseconds. An observed rise time of 6 seconds is, naturally, significantly longer, and just seemed a bit odd. I know where its occurring in my circuit, but was just trying to get a feel from AD, if that 'extreme' settle time is still OK.

    To clarify, the VREF+/- aren't starting at -7 and -8V, respectively. The scope shows a different axis for each plot. In this case, the green (VIN-) is layered on top of the black (VIN+). In both cases, those plots use the 2 vertical axes on far left of the plot, and it can be seen that they are at zero.

    Thanks again.