I've built the first prototype for an amplifier with adjustable gain of 13.000 to 130.000. The circuit works fine (even when it's not a piece of art). The problem I'm facing now is I have a lot of noise in the input which I'm able to remove by twisting the sensor cable. I put the oscilloscope in the out put and start moving the sensor cable until noise goes from several volts to under 1Vpp.

My question: Is there any procedure for doing this?

I know I can put this inside a case and work with the cable until noise goes down. But I could be missing something obvious. Should I change the flat cable for a meshed one? (Sensor output is 100 to 2500 Hz)

 

U need to use shielded Audio cable to connect low level audio signals mate.

By the way, the shield should be connected to ground.

 

A microphone produces a signal level of only 5mV with low level music or speech and the connecting cable must not produce a noise signal higher than 5uV. So shielded audio cable is used. For extremely low noise pickup by the cable then a balanced line is shielded. The balancing cancels any noise that leaks through the shield.

Even when one audio product feeds a 0.5V signal to another audio product, shielded audio cable is used to reduce the noise picked up by the cable to less than 20uV.

Telephone lines do not produce 50Hz or 60Hz so mains hum pickup is not much of a problem. The cable has twisted pairs and are balanced to reduce most noise pickup (mostly crosstalk) somewhat.

 

If you use shielded twisted pair for the cable, the shield should be grounded on one end to prevent a ground loop; ground at the end that yields the best noise performance. If your amplifier is configured as a differential amplifier it can reject residual pickup or ground injected noise, but the diff amp BW much be greater than the noise BW.

 

What signal is used for the sensor? What voltage and currents? Schematic?

 

I will try with a simple two-conductors shielded cable tomorrow. The shield will be the actual ground (as I have Vcc, GND and IF connectors). If it doesn't work I will get another cable with at leas three conductors inside so I can connect the shield to one end only.
Unfortunately I don't have a sensor datasheet with such level of detail. But I know empirically that I can reduce the noise by twisting the cable. So there should be a way of getting the noise level low enough using a proper isolation.

 

Twisting the wires allows the ground wire to partially shield the signal wire.
But an audio shielded cable has the ground wire as the shield that almost completely shields the signal wire. A huge difference.
Some shielded audio cable has foil as the shield that is a perfect shield (but the cable is not very flexible). Flexible shielded audio cable uses many braided wires as the shield that is flexible but is not a perfect shield.

I bought some expensive shielded audio cables at RadioShack. The "shield" was only a few thin wires that shielded very poorly then they broke.
Now RadioShack is gone from Canada, guess why?

 

Great! The shielded cable works almost perfectly. Way better than the flat cable, of course. Noise went down to under 1Vpp which is perfectly fine for this. As the processing unit only has to count frequency only, and has adjustable trigger level. Which is set to pulses over 2V usually.

The cable I've used has a ground conductor with an aluminum film as shield for the other two conductors. The picture was taken from another connector (a plug), not the one I've used in the actual board. There I had to adapt the flat cable connector to the shielded cable. I did that as near as possible to the board. In the following versions of the board I'm going to use a stereo plug, I guess.

 

Ground the shield at one end.