This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

INA350: When measuring an EMG signal, only 50 Hz noise

Part Number: INA350
Other Parts Discussed in Thread: TLV316, INA128

The sensor uses a unipolar power supply, but thanks to a high-pass filter, the signal is shifted to the midpoint (Ref = 3.3 V / 2 = 1.65 V). At the output of the amplifier, there is a notch filter with a frequency of 50 Hz. However, the output contains only noise with amplitudes up to mV. What could be the error in my design?

  • Hello Michael,

    Is there an input source during the output check?

    Is there something off this schematic that generates 1.65V for VREF. Is REF a good 1.65V?

    Is the noise same or similar with power off. Most oscilloscopes with a 10X probe and wide bandwidth setting can show this mush noise with a no noise input source (noise is internal to oscilloscope).

  • Yes, the input signal is taken from EMG electrodes. The reference voltage is generated by a voltage divider through a repeater on TLV316, with the power off, the noise is slightly less, generally dominated by 50 Hz, but I think this is due to the noise amplification through INA. However, the EMG signal itself is not visible at the output. The 3.3V power supply is provided by LDO and battery. I am not sure anymore whether it is possible to use the INA350 as a frontend for EMG sensors like the INA128:

  • Michael,

    Based on minutes of research (I'm not an expert) EMG signals can be very small 50uV. Which gain INA350 you are using?

    Does your input filter pass the frequencies desired?

    How our you measuring the output and noise? You made need to make adjustments or add another low noise preamplifier in between INA circuit and your measuring device.

    INA128 is configurable for higher gain, higher bandwidth, and lower noise than INA350. However INA350 should be able to handle this depending or your requirements for frequency range.  

  • All input filters are designed to cut the frequency of 50 Hz. The signal generator has checked that a signal with a frequency of 80 Hz or more is correctly amplified on the Instrumentation Amplifier (gain always 50) and on the subsequent non-inverting amplifier based on TLV316. However, when testing with EMG electrodes, there is no signal, all that is available with mV appliqués is noise at a frequency of 50 Hz (I am sure that this is noise because if you move away from the 220V cables, it decreases significantly). At the same time, the input impedance of the INA350 should allow it to be used as a frontend for measuring biopotentials

  • Michael,

    With the signal generator did you also verify that 50Hz is strongly attenuated? The length in probe wire can make a decent 50Hz antenna so that part makes sense. Are the EMG electrodes proven good? Is AC coupling a valid way to read the probes? 

    You proved the amplifier works, the interface to the probes may be the issue. 

  • Thus, using the UNI-T UTG2062 signal generator, at a frequency of 50 Hz, with an input signal amplitude of 1 mV, the output amplitude (after a total amplification of 550 times, where the amplification is 50 times from the INA350, and 11 times from the non-inverting amplifier) is about 10 mV. Perhaps the problem is in the faulty electrodes, although they are ringing. In this case, the EMG is variable, so the DC component is cut off (since it can be a parasitic galvanic element between the skin, sweat, and electrode)

  • Hi Michael,

    I am relatively unfamiliar with EMG. Is there any way to evaluate with different electrodes? It looks like the 50Hz filter is doing its job well, this makes me wonder if the signal from the electrode is not what the amplifier expecting. 

    INA350 had very high input impedance, and relatively low noise, so this should be possible with INA350.

    What is the frequency range used in this application?



  • All the filters were designed according to the expected frequencies (75-500 Hz), unfortunately, at 70 Hz there is still some attenuation from the notch filter, but from 100 Hz there is almost no attenuation. I have already ordered new electrodes, so I hope that the situation will improve

  • Thanks for the details on the passband frequency.

    Please let us know if the new electrodes improve the performance. I hope this will give us a usable output signal.