RMS Detection is Frequency Dependent?

I'm trying to sense a sudden change in impedance with the use of a LTC6268 transimpedance amplifier and an RMS averaging component. So far I've tried the ADL5511 and AD5904. These both work as intended, however only at frequencies upwards of 10MHz. I'm sending 300kHz across my input impedance (~30k ohms) with a ~50ohm variance and into the transimpedance amplifier which then feeds to the truPWR RMS detector. Basically, the thought process is as follows: measure the AC current across the variable impedance, convert it to an AC voltage with the transimpedance amp, then convert it to a DC signal with the RMS detector. I'll monitor the output of the RMS detector for a sudden change. But, I'm having trouble detecting the dip with the RMS detector at anything less than 10MHz with either board. Modifications were made to the ADL5511 in order to reduce the frequency with no improvement. Neither of the RMS averaging chips respond with a changing impedance at 300kHz. But, when running at 25MHz my impedance change shows up beautifully, and becomes even more apparent at higher frequencies. RMS value should be independent of frequency.  What am I missing here? Why do these boards require such high frequencies to see changes in impedance?

  • 0
    •  Analog Employees 
    on Sep 19, 2018 11:18 PM over 2 years ago

    This is not my usual beat, but if I may jump in, I think the two parts mentioned are not spec'd for frequencies below 10MHz in the case of AD5904 and 100MHz for ADL5511 and thus might not function well down there. Although both datasheets say "DC to 6 GHz," the truth is that "DC" is a relative term, and for RF it could easily be 10-100MHz.

    The best clue to figuring out the lowest guaranteed operating frequency of a part is to see what frequency the specs and plots in the datasheet begin at, and AD5904's datasheet plots and specs start at 10MHz, while ADL5511's datasheet plots and specs start at 100MHz.

    To detect 300kHz RMS, I would recommend a lower-bandwidth part like LTC1968: http://www.analog.com/en/products/ltc1968.html Theoretically LTC1968 goes to 15MHz, but I'd personally stick to <500kHz for this part given the input signal bandwidth vs output voltage plot on p. 5 of the datasheet and the falling CMRR past 100kHz.

  • +1
    •  Analog Employees 
    on Sep 22, 2018 1:48 AM over 2 years ago

    Hi epietrowicz,

    For extending low-frequency capability, you might want to try larger detector input coupling capacitors on your ADL5511 board, for example C1= C2= 10nF. Increase C17 if output ripple is too high.