Removing 50/60Hz noise

Hi all,

I'm having a bit of trouble removing some noise from my circuit coming (somehow) from the mains power.

The device I've designed is portable and to be used outside, so I don't expect the noise it to be an issue with the final design but for lab-based testing I'd still like to eliminate it.

The circuit is just a basic TIA/photodiode measuring a LED light (modulated slowly at 0.2Hz) which is driven from the Arduino. The light levels are very low - such that I'm getting a few hundreds of picoamps of photocurrent. I figure that this is where the noise is getting coupled into the system (this photocurrent is driven through a gain resistor of 100Mohms). I've tried a few things to remove it, but it is proving to be quite elusive. Looking at the signal with the CRO I don't see any modulation at 0.2Hz, but instead a larger 50Hz peak every time the LED is on, which then decreases in size when the LED switches off. This doesn't make sense to me - I would have thought I'd see a DC increase in the level with the LED on and the 50Hz noise being the same regardless if the LED is on or off.

The lasted attempt involving putting the whole setup (Arduino, TIA circuit & batteries) within a diecast aluminium box, and run the photodiode through coax. I grounded the circuit to the metal case, such that the anode of the photodiode (via the coax) shares this ground.

Is this the best bet? Next step will involve taking the whole setup to a complete new area to see if is location specific.

Any suggestions would be much appreciated.


Some preliminary questions popped into my head -

  1. are you powering this from an AC source? Will it go away if you powered it from a battery?
  2. is it possible that the photo diode is picking up something at 50Hz and introducing it into your circuit? Maybe fluorescent lighting?

Hi Rokkit,

The circuit is stand-alone and powered by battery. The only thing connected to AC is the CRO itself. Even when in the dark (away from any potential optical sources of 50Hz) I still observe the coupling.

Is it best to have the GND/COMM connected to the metal housing?

Yes, ground the case.

Since the bandwidth you need is very low, you can add plenty of low pass filtering.

I've considered filtering but as the 50Hz noise isn't expected to be present in environment in which the device is operating, I'd prefer to avoid adding further stages to the circuit. I'm more interested in understanding how the noise is getting coupled in.

Are there any other shielding tricks I could try?

I'm having the same issue with my simple load resistor + photodiode circuit. It's not the light source, since the noise is still present when a battery-powered light is presented to the diode.

Perhaps the arduino's ADC is not a good platform for detector circuitry :~.

With 100M impedance you have to shield everything assiduously, just bringing your
hand near the circuit will couple significant voltage.