LPF before or after the opamp?

Discussion in 'The Projects Forum' started by nickagian, Jun 2, 2011.

  1. nickagian

    Thread Starter Active Member

    Mar 12, 2010
    34
    0
    The general concept is the following: I have the HTG3533 humidity sensor and want to interface it with an ADC that has an internal voltage reference of 1.2V

    What I plan to do is to use a voltage divider, consisting of two 1MΩ resistors. Thus, the output voltage of the sensor will be divided by 2 and be always in the input range of the ADC. After the voltage divider, I'll put a buffer constructed around an LPV511 OpAmp in order to provide small output resistance for the connection to the ADC.

    Moreover, I would like to put an RC low-pass filter (something like a 5k with a 1 or 10u) in order to filter out any noise at the output line of the sensor, since the sensor will be located at a distance from the interface PCB.

    My question is : Where to put this filter? Before the resistive network, just at the entrance of the signal at the PCB or after the buffer and before the ADC? I think that the best is to put it before the voltage divider. Is this correct?

    I also have a (quite silly) question with this design: the data-sheet of the sensor says that its output impedance is 50Ω. Is there any problem that this small resistance is going to be connected to a huge input resistance of the analogue front-end? (As I said, I have a voltage divider consisting of two 1MΩ resistors. Their value can of course be reduced)

    Thanks
    Nikos
     
  2. wayneh

    Expert

    Sep 9, 2010
    12,150
    3,058
    I won't pretend to fully understand your design without seeing a schematic, but I think it makes sense to filter at the latest possible step, to include filtering noise from as many sources as possible.
    Again without fully understanding things, I'm guessing this is indeed a problem. Ultimately you're detecting a voltage drop across the sensor, by flowing a small current through it? This setup makes any voltage across the sensor a drop in the ocean compared to the large drop at low current across your divider. A link to the sensor datasheet would be nice - I couldn't find it right off.
     
  3. #12

    Expert

    Nov 30, 2010
    16,337
    6,821
    My first impression is different. Use the resistors in your voltage divider as input impedance for the low pass filter. You should probably drop to 100k or 47k to be sure the leakage of the capacitor doesn't foul the DC reading.

    That's as far as I go without seeing the schematic and the parts specs because I might be wrong already!
     
  4. nickagian

    Thread Starter Active Member

    Mar 12, 2010
    34
    0
    I'm sorry that I hadn't include any schematic in my previous post. I was in a hurry before. And what's more, I don't currently have any schematic editor application in this PC.

    I have drawn something draft below.

    @wayneh
    This sensor directly outputs a DC voltage. I'm sorry for the misunderstanding and for not uploading the datasheet. I provide a voltage and ground to it and it gives out a ratiometric output. Just, it's range is above the 0-1.2V input range of the ADC. That's why I need the voltage divider. Regarding your recommendation of putting the LPF after the buffer. In that case, the output resistance before the ADC will not be that low comparing with using directly the output of the opamp, isn't that correct?
     
  5. #12

    Expert

    Nov 30, 2010
    16,337
    6,821
    If you moved the capacitor to the junction of the (2) 1M resistors, you could get the same effectiveness with .1uf and use a very low leakage film capacitor.

    ps, I don't mean select a low leakage cap, I mean film caps inherently have low leakage.
     
    Last edited: Jun 2, 2011
  6. nickagian

    Thread Starter Active Member

    Mar 12, 2010
    34
    0
    Thank you for your suggestion! I haven't thought of that!

    By the way... do you feel that the cut-off frequency of this configuration (~3Hz) is more than enough? Perhaps I could go a little higher at ~30Hz by using a 0.01u instead.
     
  7. ErnieM

    AAC Fanatic!

    Apr 24, 2011
    7,394
    1,606
    Wow, it really IS 3 Hz, its impressive you got that as most people would not see the cap breaks with 500K.

    I would keep it at 3 Hz, give yourself all the room away from power line interference you can.

    How far apart is the sensor from the A2D? Inches or feet? I'm asking as I think the meg ohm resistors may cause a problem, just high impeadance points are better at capturing noise then low z ones. If it does you may need that 5K again on the board from sensor input to ground.
     
  8. nickagian

    Thread Starter Active Member

    Mar 12, 2010
    34
    0
    :) Yes. The equivalent resistance of this resistive divider is found by taking the two resistors in parallel (1M // 1M).

    Oh I see.. you mean to put a 5K from the active line to ground in order to minimize the noise? Actually the distance of the sensor from the A/D will be something between 1m and 50cm. Or even 30cm. But in no case shorter. Thus, I guess in the range of feet.

    It's interesting.. you know. Things like that, putting the low impedance at the entrance of the signal on the pcb for lower noise, is sth that I have never learned or read in any book during university. Or even now. I mean, in the end, practical experience is really important!

    Thanks for your advice!
     
Loading...