Sanket:

Low impedance in electrodes used for biopotential recordings such as EEG and EMG is required to provide impedance matching with the environment you are measuring from so as to maximize the power transfer otherwise reflections by the load might occur - a bottleneck effect.

Now, the reason the kilo-Ohm is preferred for the input impedance is so as to keep the current in the circuit in the milli-Ampere range for electric safety.

Thank you for the clarification. I am conducting a research with a neurologist regarding the same. He wants to look at the differences in the signal pattern for variable values of the impedance. The current value used in India is very high and I am conducting this research to generate awareness amongst other neurologists here.

You are most welcome Sanket!

Hello, Sanket, I would like to know the results of your work upon impedance. Please post the article here when it is published.

Higher impedance results in higher noise. Unless one requires high impedance electrodes (for example for single neurons), lower impedance gives a cleaner signal.

Further to Parveen's answer it is also important that the impedances are matched between the 2 electrodes (and the ground).

No... The reason for low impedance is because a high impedance input would be in parallel with the source impedance of the skin (typically about 150K). The signal will find the path of least resistance. If you were to use a high impedance input, the signal would simply travel through the skin instead of into your electrodes.

Sam, I am not sure what your answer means. The electrical signals are flowing all the time to activate the muscle. When muscles are active, currents flow in every direction including the skin due to volume conduction. We want to detect these currents. The measured voltage resulting from these currents will be determined by the properties of the electrodes you use. In other words question is what is the best way to record these electrical currents of individual muscle fibres, a single muscle or a population of muscles. Surface electrodes with low impedance provide us with a population activity with least noise. Low impedance electrodes with cleaned skin (decrease skin impedance) allow larger current flow from the source to the electrodes. At a particular position on the skin the amount of current flowing towards it is the same irrespective of the electrode properties. How much voltage you measure depends on the properties of the electrodes. Since there is sufficient current from a population of muscles, even low impedance can show measurable voltage. They pick up noise from other sources, but because of their low impedance signal to noise ratio is good. But the high frequency component in these low-impedance electrodes is reduced. High impedance electrodes, such as intramuscular wires, provide selective signal from either one muscle or a few motor units. These high impedance electrodes are used when one has to measure signals from small current sources. Though these electrodes provide measurable voltages of small currents, they are prone to picking noise (movement artifact, line voltages 50-60 Hz).

Over the last three years I have developed some pretty advanced EMG sensors.

It is important to be clear in a discussion like this. Current does not flow in every direction. Also, I am not necessarily explaining this to you (as you appear to be quite knowledgeable but to someone else that may read this in the future).

Current flows from a higher potential to a lower potential - the body is no exception. The brain generates the voltage to drive the signal in the appropriate nerve to the muscle being driven. The primary direction of the current flow is in the axial direction of the muscle. All other flow is so much less, it is inconsequential.

Moreover, technically flow is not taking place all the time - even for a single activation cycle). This is an AC signal so the brain is raising and lowering voltage potential - pumping electrons forward and backward at rates ranging from about 70 Hz to 400 Hz throughout a given muscle activation depending on activation intensity. So even during activation, current flow stops dozens or even hundreds of times.

As the signal travels axially through the muscle, some current leakage takes place radially to the surface of the skin above the muscle (Albeit with very little voltage potential behind it) if a current path is available - say into an electrode.

This happens more easily as sweat ducts fill with electrolyte solution. Since the signal travels primarily axially through the muscle, a voltage potential will reach the first electrode before it reaches the second. It is THIS difference you are picking up with the differential amplifier. In fact, what we are really picking up, is the phase amplitude difference of the wave being transmitted by the brain. It is therefore somewhat random what the amplitude will be for a particular wave period due to the propagation delay through the body and the electrode separation.

If the Amplifier has a low impedance input and the electrodes are high impedance you could have a problem because you would have a voltage divider formed by the sweat on the skin in parallel with the impedance of the electrodes themselves. This is the reason for the reduced signal, and the point I was making.

For ambient EMI rejection, the most important factor for good signal integrity is good common mode rejection.

Why do you believe that a low impedance electrode necessarily reduces high frequency components more than low frequency components? This stuff is ALL low frequency. Electrode impedance should have nothing to do with this. The only reason for measurable frequency attenuation would necessarily have to be a frequency filter. We aren't talking about Megahertz here. We are talking about a few Hz.

To be clear, I don't care for low impedance sensor systems. The voltage divider I describe above is one reason why. They attenuate the kill the signal far too much.

A high impedance electrode is far superior as long as it's well designed to both shield against and common mode noise away. I never bother preparing skin. Most of the time I get an exceptional signal putting it on unclean skin and I certainly don't shave my arms when I use the sensors on myself. It's all about common mode rejection.

A good high impedance differential amplifier has input impedance in the teraohms. A few hairs aren't going to present much resistance that matters.

Sam

It is simply a problem of power transfer from the source (brain, muscle, etc) to the first stage of pre-amp.

If your amp is high impedance, say in the hundreds of megohms to gigaohm range -- then skin impedance is less important. See this link.

Article Scalp Electrode Impedance, Infection Risk, and EEG Data Quality

Here's another related paper, which cites the previous one I mentioned.

Article The Effects of Electrode Impedance on Data Quality and Stati...

The periodicity of the signal is low frequency, but the frequency spectrum of what might be approximated by a delta function is quite broad and the frequency response of the measurement system could distort the waveform.

The electrode impendance is a serial circuit to the source (brain) and the recording system. Every impedance produces noise that is proportional to the Impedance. If the impedance is 5kHz, the impedance noise is too large, you can not measure the brain activity any more. See also my page for a more detailed explanation, with example spectra

https://pub.ist.ac.at/~schloegl/qc/

Just for clarification, I was referring to EEG. The question was about EMG/EEG, EMG and EEG are very different becase EMG has much larger amplitudes (milliVolt), while EEG is in the 10s of uV range. Therefore, a low impendance is much more important for EEG recordings. A 5kOhm impedance over 100Hz bandwidth produces almost 0.1uV rms, 4 four times the impedance or bandwidth, would double the noise amplitude.

Johnson Noise. AKA KT noise, thermal noise, and Nyquist noise, in an amplified, high-impedance circuit is one culprit. This is caused by thermal fluctuations within conductors and resistors. The voltage of KT-noise: V = [4 K_B T R f_bandwidth]1/2, where K_B is the Boltzman constant, T is the temperature in Kelvin, R the resistance in Ohms, and f_bandwidth the frequency bandwidth of the circuit. One can look at this relationship and readily see that the higher the resistance the higher the noise for a given frequency band and temperature. Let's say that your circuit is at room temp or near it, T = 300K, the circuit impedance of 100 Mohms, the frequency bandwidth is in the 1.0 GHz area. V = 40mV of noise. Might need some bypass capacitors. Another source is shot noise. This noise is due to current flow through different electrical contacts or regions in a high-impedance circuit. Solder joints have dissimilar metals therefore they have a thermocouple-type effect, circuit conductivity transitions like an op-amp from the conductive inputs to the active semiconductor, etc. Shot-noise is also temperature-dependent because of Ohm's Law and the circuit resistance temperature dependence. I_noise = [2 e- I f]1/2 where e- is the charge on the electron, I is current, f is the frequency bandwidth in Hertz. V_noise = I_noise R, R is the resistance of the circuit in Ohms. R is also dependent on temperature! I_noise = [2 (1.602E-19C)(1 microAmp)(1GHz)]1/2 I_noise ~= 20nA or 2% of the total current. Note it is current dependent. V_noise = (100MOhms)(20nA) V_noise = 2V. The applied voltage to the circuit is: V_applied = (1.0microAmps)(100MOhms) V_applied = 100V. So the shot noise is 2% (roughly) of the applied voltage. These are rather rough calculations and rough explanations. Still, they are "close enough." Which noise dominates your particular circuit depends on the circuit!