-1

I have got a analog signal (from pH probe) that include some unwanted AC noise. The noise is about to 0.5 mV.

I want to use passive RC low-pass filter but I'm not sure which frequency that I need to pass. When I use 100 nF cap. and 100 kOhm resistor (it means it allows below approximately 15 HZ ?) in front of ADC, I see some offset voltage around 150 mV.

How can I decide the frequency of the noise or frequencies that need to be allowed?

Also do passive RC low-pass filters always put some offset voltage to the signal?

user30878
  • 185
  • 17
  • 1
    English sentences start with capital letters. We deserve more respect than this. – Olin Lathrop Nov 04 '13 at 12:56
  • 1
    The 3 dB "knee point" of an RC filter is 1/(2*pi*R*C) Hz. After that, it will fall off at 6 dB/octave (doubling in frequency.) The "DC" offset is likely just the result of filtering higher-frequency components, or it's a result of poor de-coupling between filter and ADC (use an opamp to re-buffer.) – Jon Watte Nov 04 '13 at 16:40

2 Answers2

2

Do you know your analog signal's frequency(ies)? That is the frequency(ies) you need to pass. The ac noise could be below that or beyond that.

Also use an opamp as voltage follower after the RC filter in order to prevent loading effects to the filter, and thus altering its frequency. opamp voltage follower

1p2r3k4t
  • 696
  • 2
  • 9
  • 18
2

Most ADCs that are built into a microprocessor (like a PIC or an ATtiny) require that the signal impedance is low and this is fine because you are using a 100nF capacitor BUT, leakage currents from the ADC's input mean that the signal DC resistance must also be low or a DC error term is created.

With 100k ohm this is likely to be the answer. I know that on an ATtiny85 the specified source resistance is "less than 10kohms". Maybe use 10k ohm and 1uF instead of 100k and 100nF.

A lot of stand-alone ADCs have the same (or similar issues).

Andy aka
  • 434,556
  • 28
  • 351
  • 777
  • What about 100 ohm res. and 100uf cap.? – user30878 Mar 13 '15 at 12:56
  • @user30878 You've accepted an answer to use an op-amp - did you try this method or did you find that it gave dc errors or other problems? – Andy aka Mar 13 '15 at 13:45
  • Yes. DC offset error is erased with op-amp. But there are still some DC noise on signal. – user30878 Mar 13 '15 at 14:20
  • I need to ask capacitor type also. It won't be electrolytic (polarized) right? Which capacitor is good to use? – user30878 Mar 18 '15 at 21:49
  • Presumably you are no longer considering an op-amp as the right answer hence your question. If so please explain why. If you are considering the op-amp to be the correct approach why are you asking me this question? – Andy aka Mar 18 '15 at 21:54
  • Op-amp is the right answer, as I said yes before, DC offset error is gone with active filter configuration. But still some DC noise on the signal remains. So I think its about leakeage currents and I want to decrease resistor value as you advice me. – user30878 Mar 18 '15 at 22:08
  • Just what sort of DC error were you getting with a 100 ohm resistor and also what op-amp are you using? What about its offset errors? Also what is the signal that feeds the op-amp in terms of voltage level range and its output impedance. In case you were wondering, I don't think using an op-amp is at all justified for this. – Andy aka Mar 18 '15 at 22:18
  • I use LMC6042 for filtering. I didn't try with 100 ohm or 10k res. still, but I will try and write the results. I tried with 100k. The signal voltage range is between 1-3 V. This signal is coming from output of another LMC6042 (U1) op-amp in non-inverting mode with 0 gain(so impedance isn't problem in here?). When I use pcb. adapter for connecting U1 op-amp(SO8 package) the result is [this](http://i.imgur.com/MlGUtja.png) . But when I use [this](http://media.digikey.com/photos/Logical%20Systems%20Photos/PA8SO1-03-3.jpg) SOIC to DIP adapter noise decrases ten fold. – user30878 Mar 18 '15 at 23:41
  • Sorry I deleted and wrote again the message above. – user30878 Mar 18 '15 at 23:47
  • If you are feeding from another 6042 then why have this extra op-amp. It doesn't make sense. Try 100 ohm or maybe 1k ohm and see what you get. BTW the graph is meaningless without scales. – Andy aka Mar 19 '15 at 08:09
  • You are aware that the biggest error is likely the input offset voltage of 3mV for the op-amp and if there are any further upstream that have significant gain then this 3mV is multiplied by gain. – Andy aka Mar 19 '15 at 08:16