How to remove 60Hz noise from square wave?

[Topher925]

I have an amplification circuit where I am essentially reading a somewhat deformed square wave and I am having a lot of trouble with 60Hz noise from the walls being superimposed on my signal. The signal of interest goes from 5v - 0v, has a period of about 60ms, and I need it to be relatively unaffected by any filtering. I've tried low pass filters and while they get rid of the 60Hz they also attenuate my signal. Notch and bandpass filters just seem to differentiate my signal or completely distort it. Is there a good (preferably simple) way of removing this 60Hz that I haven't tried? I'm not an EE and only know the very basics of circuit design and unfortunately this problem seems to be over my head.

 

[f95toli]

No, there is no good way to get rid of 60Hz once its in your system (in theory you should be able to use a very sharp notch filter, but that never works well in practice).
The best way to avoid problems is careful grounding (a ground-loop will inevitably lead to 60 Hz noise) and -when working with small signals- battery operated pre-amplifers; i.e. avoid getting the 60Hz in there to start with.

There are a few "tricks" one can use, e.g. synchoronizing the measurement with the mains (which is how some benchtop multimeters avoid 60Hz interference).

Do you know where the 60Hz is coming from? Is it thru the cables or is the problem inductive pickup?

 

[Topher925]

No, there is no good way to get rid of 60Hz once its in your system (in theory you should be able to use a very sharp notch filter, but that never works well in practice).

Thanks, now I don't feel quite so dumb. My source of noise is from a photodiode, which is also the source of my signal. The thing is in a steel housing with an aluminum ST connector which is acting as a big antenna. I have the thing grounded of course but there is still a about a 12mV 60Hz sin wave in the raw signal.

When you talk about "synchronizing with the mains", are you referring to common mode rejection? Both the opamps in my circuit have relatively high CMRRs which I have tried to exploit but so far, no dice.

 

[Bob S]

12 mV 60 Hz noise is very high, and careful grounding and shielding of all cables should eliminate this. There should be only one common ground to the house ground. All other grounds should be connected to this one. secondly, no ground loops. Make sure there are minimal multiple paths for connections between chassis and instruments, microphones, etc. where Faraday induced voltages and currents can be introduced. Ideally both wires from the photodiode be floating, and one side connected to ground only at the amplifier, and nowhere else. Transformer stray magnetic fields can also induce 60 Hz. A photodiode itself should not be a source of 60 Hz noise, unless it is on the light signal itself.
Can your signal of interest be a sub-harmonic of 60 Hz (66.66 ms) and synchronized to the 60 Hz line frequency?

 

[mheslep]

I want to 2nd f95toli's post for emphasis as the OP's instinct seems to be a common first approach: expending effort on filtering out 60Hz after its introduced to the system. Don't do it. It is almost always more productive spend effort on stopping the introduction of 60Hz into the system in the first place through better grounding practice or materials.

 

[Topher925]

A photodiode itself should not be a source of 60 Hz noise, unless it is on the light signal itself.

But it most definitely is the main source of noise. I have it grounded correctly just as you say and what how the manufacturer (Hamamatsu) suggests. If I put a cap across the photodiode or ground the positive input then all of the noise goes away. The only other thing I can think of is to put the thing in a metal box or build a filter.

 

[dlgoff]

A good rule of thumb that I use is: get rid of the noise at the source. Which may be hard if you have to have the mains intergized.

What light source are you looking at with the Hamamatsu? Could your room lighting (flouresent perhaps) be the source of the 60Hz (as Bob S says: "A photodiode itself should not be a source of 60 Hz noise, unless it is on the light signal itself.")?

 

[berkeman]

Can your signal of interest be a sub-harmonic of 60 Hz (66.66 ms) and synchronized to the 60 Hz line frequency?

Small typo -- 60Hz has a period of 16.67ms.

 

[berkeman]

But it most definitely is the main source of noise. I have it grounded correctly just as you say and what how the manufacturer (Hamamatsu) suggests. If I put a cap across the photodiode or ground the positive input then all of the noise goes away. The only other thing I can think of is to put the thing in a metal box or build a filter.

If the 60Hz noise goes away when you put a cap over the photodiode, then you are picking up optical light 60Hz. Can you turn out the lights?

(Also, noise from the lights should be 120Hz, rather than 60Hz...)

 

[Bob S]

Bob S said
Can your signal of interest be a sub-harmonic of 60 Hz (66.66 ms) and synchronized to the 60 Hz line frequency?

Small typo -- 60Hz has a period of 16.67ms.

Actually not. The OP mentioned 60 ms period. I was suggesting the fourth subharmonic of 60 Hz (= 15 Hz, or 66.67 ms period).
Bob S

 

[berkeman]

Bob S said
Can your signal of interest be a sub-harmonic of 60 Hz (66.66 ms) and synchronized to the 60 Hz line frequency?

Actually not. The OP mentioned 60 ms period. I was suggesting the fourth subharmonic of 60 Hz (= 15 Hz, or 66.67 ms period).
Bob S

Ah, sorry. My mistake. Must stop skimming posts...

 

[Topher925]

I finally figured out the source of noise this morning. It was indeed EMI and not the lights in the room. It was coming from the steel base plate acting as an antenna that my breadboards are mounted on. I noticed it when I touched a bare part of the plate and the noise in the circuit jumped. After I grounded the circuit to the plate with a screw the 60hz disappeared and now the only noise I have is a ~50mV random noise in the final signal which I probably won't need to get rid of. Any capacitor or filter I add just makes the noise worse so I'll probably just throw the final design in a metal box and call it done. I feel kinda stupid for not figuring it out earlier but at least now I know better.

 

[Cyrus]

Hooray for DAQ systems, the pain in everyones ***!