AD8436 long startup time

I am finding that the AD8436 RMS-to-DC has a significant startup behaviour - 20+ seconds for my application.

I am using the AD8436 evaluation board: +/- 5 V supplies, IBUF disabled & bypassed, DC coupled, OBUF enabled.  I need to measure pulsed signals down to 5 Hz with crest factor > 6, so have added 100uF (tantalum, 6.3V, Kemet T491) to the existing 10uF to achieve the required RMS conversion accuracy.  Other components as standard.

After applying the supply power, the output voltage is zero regardless of the input signal until some amount of "startup time" has passed .  After the startup time has passed, the output settles to the correct value according to a decay as exemplified in Figure 30 of the datasheet.  The startup time seems to vary roughly in proportion with the value of Cavg and in inverse proportion to the input signal applied during startup.  For example, for a 20 mV input, the startup time is around 20 seconds.  For a 5 mV input signal, the startup time is 70 seconds.

In my application, the input signal will be < 2 mV for 10 seconds while the system boots up, by which time I need the RMS converter output to indicate the correct RMS value.  Under these conditions, the AD8436 will still be "starting up".

Currently, I am thinking that I will have to inject some current into the SUM pin for a second or two after power on to bring the startup time below 10 seconds.

Please could you comment:

  1. Is this the expected behaviour, or is something wrong?
  2. If this is normal, is there some other way to reduce the startup time?

Thanks in advance,

Douglas Carter

  • Hi Douglas,

    Thanks for your email and feedback. Yes we are aware of the slow startup at turn on with 0 or low input voltages. You are quite perceptive in that a higher input current will help and I've addressed this in a recently released application note, attached (see pages 8-9). You will see that the device is scaled so as to behave as before (i.e. unity gain transfer) and the input resistors are scaled to 1/10 the original values. There is a tradeoffs however (isn't there always?), specifically that (1) the device will exhibit additional error at higher input voltages and (2) the precise temperature inherent with the two internal resistors are bypassed and temperature tracking will depend on the properties of the external resistors.

    We have tested a better method but it requires two additional PNP transistors and a few resistors. If you're up for it, I can send you a schematic and tell you how it works. In this case the tradeoffs are additional external parts. This is no 'fix' for the faint of heart but It sounds like you have the knowledge and insight to use it successfully.

    I think you can  click on a link to get my email so let me know which option you'd prefer.

    Best,

    James - AD8436 apps engineer

  • Hi James,

    Thank you for your reply.  I can't seem to find your direct email address.  Please let me know how to get your email address, or post details for your PNP solution.

    Best regards,

    Douglas

  • Hi James,

    We are using AD8436 on our new product now, and having this long settling time issue at low AC input (1mV). Could you point me how to use transistors to inject current as you mentioned above? Thanks in advanced.

    Best Regards,

    Cindy

  • This question has been assumed as answered either offline via email or with a multi-part answer. This question has now been closed out. If you have an inquiry related to this topic please post a new question in the applicable product forum.

    Thank you,
    EZ Admin