Hello Namig,

maybe you could show the schematics and the pcb-layout. I used to have problems with my grounding concept (with an OPA627). I achieved better results by providing analog GND, analog V+ and V-, and some snall capacitors on the output line (I measure sunlight, so no high frequencies to care about). What is your application? If you provide more details, you will get more helpful answers.

Cheers

Sebastian

hello Michael,

thank you be interested with my problem.I added circuit.first stage is I/V converter second stage and third stage are filters( HPF and LPF).main problem I gueessed is about supply voltages.I use OPA 627 in filter stage. I can see 50-60mV signal on oscilloscope when input circuir is not connected source.application is detecting modulated signal.bandwidth should be 10-100khz. can you explain how you use analog V+ and V-.I guess noise comes from my power supply also. I use old power supply.for overcoming I put two caps to legs of power supply.Can you send schematic that you used?

sincerely

Namig

I don`t know at which node you have measured with the scope. However, your filter circuits look somewhat "strange". Your passive highpass is heavily loaded and your active lowpass has a filter Q of 0.5 only. 

Hello Namig,

by analog V+ and V- I mean that I provide only filtered power to my opamps. I added a pdf by adafruit, because there the filtered power is achieved the same way I do it: just put a ferrite bead between the power supply and your V+ plane (and your V- plane and your GND-plane in the respective lines). This decreased noise in my setup considerably.

Good luck and tell me if it works

Sebastian

Hello Namig,

assessing noise having the input open was never a good idea as your circuit will 'catch' whatever frequencies are 'on air'. Frequencies in the tens of Kilohertz range are often emanated by fluorescent lighting so you might want to switch the lights off.

In general you better short the input to ground to avoid catching disturbances...

I agree with Sebastian that an LC filter (ferrite with reasonable inductance followed by capacitor) is good to improve the supply quality. And: you should add even larger capacitors (10 uF..100 uF) to increase supply stability.

One more issue to think about: resistors add resistor noise which is proportional to the resistor value. Using the lowest-possible values for the resistors in your input stage should also result in a reduction of noise.

Hello Lutz, 

can you suggest solution for filters?

To recommend a filter is not a problem - however, this requires a kind of specification: What do you want the filter has to do? Bandwidth, damping requirements, gain...?

 ı changed Q factor from 0,5 to 2. I changed value of C8 as 4.7 nF.  I obtained better result .thank for warning. I obtain sharper fall after 100khz,it is good for me. I want no gain filters.bandwidth is 1-100kHz. question is how can I understand "heavily loaded' for highb pass filter.Do you mean capacitor value is low ?

Hello Dreher,

thank you for suggestions. I did not know about noise oeefect of fluorescent ligthing untill you have said, I will pay importance this issue. Have you any suugestions about shielding circuit from environmental noise? I read shielding circuit  with aluminium foil and grounding this aluminium is usefull.I will add some higher capacitor at legs of op amps.

"how can I understand "heavily loaded' for highb pass filter."

Your highpass output node is R7, OK? This node is connected directly (loaded) by R4. This destroys the highpass function.

Hello Namig,

shielding with aluminum is difficult as aluminum has a non-conducting oxidized surface. Try using PCB material (eg. single-sided raw PCBs) instead, soldered together to form a box or try to get a so-called shielding case (made from tinned steel thus also solderable). And don't forget to add chokes/ferrites and capacitors wherever your lines (power as well as signal) enter the case !

During development it will be best to switch off the fluorescent lighting and use a plain old incandescent lamp - or a 'dumb' LED light without a switching power supply. You may want to set up your own LED light - operating constant current from a DC power supply to avoid interference.

so how can I connect two stages?I mean HPF to LPF

For realizing bandpass filters you always have two basic approaches:

1.) Bandpass direct realization (active or passive)

2.) Series connection of a highpass and a lowpass - decoupled with a buffer amplifier.

For selecting one of these alternatives, at first,  you have to formulate your requirements:  Bandwidth, mid frequency, damping requirements, gain...?

That`s what I have asked you already.

Bandwidth is between 1-100kHz, mid frequency 10khz.no gain required. Q =1

Are you familiar with filter theory basics? The definition of Q is Q=fo/B, in your case: Q=10/(100-1)=0.101.

Forgetting the mentioning of Q=1 your requirements can be fulfilled with a passive combination of an RC-lowpass (first stage, 100kHz) and a CR-highpass( second stage, 1kHZ). For example:

Lowpass: 100 Ohm, 16nF

Highpass: 1.6nF, 100kOhms.

Teh values are chosen so that the second stage does not constitute a considerable load to the first stage.

I agree with Mr. Lutz von Wangenheim; filter design is not really good. It would be appropriate to look into the theory of cascading filters. The passive HP filter output impedance influences the active low pass filter.