Will increase in order bessel filter in any way contribute to noise?

[raja sumant]

I am designing an eeg circuit and planning to do an adc for it. Since the eeg requires a band pass filter I am planning to use a second order low pass bessel filter in it.

Suppose I want to reduce the noise ( as I am working with low frequencies ) and increase the efficiency of the circuit, will this do or do I need to go for a better solution?

If there are any suggestions please feel free to give them to me

 

[sophiecentaur]

As a general principle, it is better not to have to deal with DC because there are many sources of low frequency noise and 'drift'. The recommended bandpass filtering would eliminate DC. It shouldn't be a great problem to high pass (AC couple) the input signal as well, if you want to use the Bessel LP design.

 

[Enthalpy]

Hey, one first subject where I disagree, although only partially, with Sophiecentaur!

As EEG signals are at low frequency, I prefer to pass the full DC up to the digital signal. Digital processing can remove the DC component and still react quickly, while analogue circuits that cut at 5Hz or 10Hz take very long to stabilize, which is irritating for real - you don't even know for sure if equilibrium has been reached, when converging from 1V unbalance to 1µV.

This needs the whole circuit to accept all offsets without saturation, so limit the gain (remember Seebeck effect), have a wide ADC dynamics and consider commercial autozero op amps. These improve LF noise anyway.

About Bessel low-pass: more poles would increase only the slope at frequencies much higher than the corner, that's what Bessel do. And far from the corner, all two-pole low-pass are identical. I wouldn't worry too much about it. If out-of-band noise hampers the conversion to digital, add some poles, but if not, filter by software.

50Hz and 60Hz is not interesting for EEG signals, is it? So instead of adding poles to the low-pass, you could put notches there. At both frequencies, so your design works everywhere - as voltmeters do. If one notch at each frequency needs a too tight tuning, put more notches.

-----

In the cases where time response is important and selectivity is necessary (EEG?) a Bessel is a bad answer. The best among the well-known functions is the inverse Chebychev, or type-II Chebychev, while an hourglass is good and an elliptic not bad.

Before telling me something else, please kindly consider that
(1) I've spent several months on that exact topic, and invented better filters for that, while book authors only copy on previous books.
(2) Response functions should be compared at identical selectivity, not at identical number of poles. Needing fewer poles, an elliptic has a far better step response than a type-I Chebychev which is less bad than a Butterworth with its many poles and resulting high Q factors.
Thank you.

To design filters, the old free FilterCAD saves much time and has built-in knowledge. It computes step responses as well. Usable without the circuits sold by LT. Download it.

 

[raja sumant]

thank you sophiecentaur and Enthalpy ... I will take your suggestions

 

[alphysicist]

Increasing the order of a Bessel filter will not directly contribute to noise in the circuit. However, it is important to note that any filter, including a Bessel filter, will introduce some amount of noise into the signal due to its filtering process. This noise may be minimal and may not affect the overall performance of the circuit, but it is important to consider when designing an EEG circuit.

In terms of reducing noise and increasing efficiency, using a second order low pass Bessel filter is a good choice for an EEG circuit. However, there may be other factors that could contribute to noise in the circuit, such as the quality of components and the layout of the circuit. It is important to carefully consider all aspects of the circuit design to minimize noise and optimize efficiency.

Additionally, there may be other filter options that could potentially provide better noise reduction and efficiency for your specific application. I would recommend researching and comparing different filter types and consulting with other experts in the field for further recommendations.

Overall, using a second order low pass Bessel filter is a good starting point for your EEG circuit, but it is important to carefully evaluate all aspects of the design to ensure optimal performance.