Help me remove noise from a EMI signal

Hey all,

I have a problem with my oscilloscope measurement. I'm building an amplifier for a microvolt signal, with about 55dB gain.

While inspecting the output of the three-stage op-amp amplifier, I can see my signal, carried on a 50Hz (or 100Hz) signal - the frequency and harmonic of mains power.

This is a problem, as the EMI signal from the power line is greater than the signal I'm trying to amplify, which causes the signal to be clamped by the op-amps.

Trying to block the signal, I enclosed the breadboard in a metal case, which reduced the interference, but it was still present.

Getting frustrated, I disconnected the probe from my circuit - only to discover that I can still see the 50Hz signal! (I get an amplitude of about 1mV). I can no longer tell whether the noise is coming from the circuit itself, or the measuring equipment. With the probe disconnected from the scope, I finally get a flat-line.

This is a show-stopper - I can't work like this, and I'm getting tired of trying and failing. This noise is just there. I am clueless as to what can I do...

Any ideas...?

Thanks...

50hz noise on the oscilloscope probe

z9u2k said:

Getting frustrated, I disconnected the probe from my circuit - only to discover that I can still see the 50Hz signal! (I get an amplitude of about 1mV). I can no longer tell whether the noise is coming from the circuit itself, or the measuring equipment. With the probe disconnected from the scope, I finally get a flat-line.

Thanks...

Click to expand...

Well a ? for u how do u see 1mV or flat line with the probe of oscilloscope disconnected. In my view your probe is giving that noise to your system as finally reference point of probe is connected to earth only.

I haven't understood your input to the system , may this possible to battery power ur board?

emi single point grounding

I don't hear anything surprizing from your post. Mains frequency interferences are simply everywhere and you have to consider them, when designing low-voltage circuits. Apparently you didn't up to now.

It's quite normal, to see some 50 Hz hum with an open probe input or an unshielded circuit.

You have already found out, that shielding of your circuit is a necessary means. The next step is to indentify the remaining entry pathes for mains interferences.
- amplifier input. It should be shorted at the amplifier without a cable connected to assure, that it's not feed to the input
- amplifier power supply
- magnetic fields, e.g. from a power supply transformer
- unsuitable shield ground connection
- interfering currents running through your circuit because of misplaced external ground and supply connections

As an additional means, check if the interference is completely/partially sourced from the first amplifier stage by shutting it down. Can give additional insights about the interference path.

emi noise handling techniques

dipnirvana said:

I haven't understood your input to the system , may this possible to battery power ur board?

Click to expand...

It is battery powered

I might have mis-described my setup. I'm actually attempting low current measurement. My first stage, before the op-amp chain, is a 2N2222 with a resistor at the collector, to buffer the signal.

I suspect that the transistor's base lead itself serves as a small antenna for the mains hum, which is then amplified by the transistor, practically overriding the measured signal. My breadboard's rail might contribute to this effect... Makes sense?

FvM said:

- amplifier input. It should be shorted at the amplifier without a cable connected to assure, that it's not feed to the input.

Click to expand...

I'm not sure I understand what you mean...

In the final PCB I'll attempt to keep the tracks as short as possible, but currently I'm unable to even prototype...

ecg amplifier hum interference shield

I'm not sure I understand what you mean...

Click to expand...

Hopefully, you understood at least some of the suggestions? You'll get the other points by thinking about the problem, I think.

Reading your second post, there are too much vague points about your setup, so I don't want to continue detailed suggestions.

Basically, you have to figure out, where the interferences enter your circuit. There are many ways to find out. It's basic engineering work, but it can't be saved by some clever hints from a distance.

Good luck

why do i get a flat line on my oscilloscope

FvM said:

Basically, you have to figure out, where the interferences enter your circuit. There are many ways to find out. It's basic engineering work, but it can't be saved by some clever hints from a distance.

Good luck

Click to expand...

Thanks

Probing a little more, I believe the interference is coming from the leads and breadboard rails. I can tell that by connecting my scope probe to the unpowered circuit and get a 50Hz/50mV sine wave at the base of the transistor. Probing other places produces the same wave, with amplitude varying from 10mV to 50mV. With the probe disconnected the signal is just 2mV, so I know it comes from the board.

Thing is, I'm not sure how to deal with that...

how to eliminate 50hz noise on low voltage signal

Having read all the post my understaning is that, in addition to the care suggested you keep in mind two important things to attend with utmost care.
1. The initial stages being sesitive contribute most. So handle them with utmost care. High impedance inputs are prone to pickups like this and the probe leads work like antennas.
2. Common mode also plays important role at relatively high current as well as high sensitivity parts. So try to follow single point ground for at least initial stages of signals.
3. Laying of wires / conductors is also to be taken care to minimise noise.
Now a word about single point ground technique. In all the sensitive (high gain) applications the signal ground from all parts is brought at one point. This is a important practice observed you can find details if neede by google search or emi literature. Not bothering about any thing (references) go ahead for single point ground, taking care to keep lead lengths low.
Role of transistor leads may not be proble in your case, such issues arise if you are handling high frequency / fast rising pulses and distortion and hf pickup is the issue.

adc emi noise

Just quick $0.02. This sounds a lot like an electrophysiology instrument (EEG, ECG/EKG, etc). In that area it's common to pick up a differential signal, transmit it through twisted pair. That way both sides of the differential signal pick up the same interference. then an instrumentation amplifier (InAmp) cancels the interference by subtracting one signal from another (and amplifying the difference). All that is done to remove 50Hz interference. Could such approach be viable in your case? Could you tell us what you're trying to measure?

single point ground technique

(note: I have changed my design a bit, and I'm measuring voltage instead of current now)

My sensor has an output voltage of about 35uVrms. It is mounted directly on the PCB. I've removed the 2N2222, so it is now connected to an in-amp through an RC filter. The amplifier is connected to an ADC.

I'm using an in-amp with high CMR (>=120dB) set to a gain of about 36dB. The signal is further amplified at the ADC for a total of 46dB.

I have an RC high-pass filter of about 240Hz between the sensor and the in-amp, which greatly reduces the 50Hz wave, but it is still too dominant.

As for CMR noise cancellation: That why I chose an in-amp with a very high CMR, but it appears I'm not picking up the signal equally on the reference and the sensor 'buses'.

The device is powered directly from a battery, so the humming can't come from the PSU or PWM (no DC/Dc convertor).

I don't know... I think I'll put another filter between the in-amp and the ADC, trying to further attune the signal.

Thanks everyone.

role of 2n2222

Rated CMR is one thing and connecting the circuit to include minimal additional common mode signal is different. The solution in short lies in useing SINGLE POINT GROUNDING. all the grounds are connected to one commond ground taking care that the leads are not too long and too thin.

how to remove the 50hz noise from oscilloscope

If you are measuring with a scope, make sure to use the ground lead on the scope probe. Not some other ground arrangement, the ground clip that comes with the probe itself. Connect that to a board ground as close as possible to the signal you are measuring. Try not to form a "loop" with the ground lead, as it will pick up any magnetic fields in the vicinity.

To do otherwise is to pick up all the noise available in the environment and totally contaminate your measurement.