Twin-T Notch Filter real applicability for 50Hz frequency suppression?

[carpenter]

I want to separate the control signal coded at a frequency of, for example, 180Hz from the mains voltage 230V 50Hz.
I use MCU ADC and Goertzel algorithm.
It works, but it wouldn't hurt to increase the weight of the 183 Hz signal, in other words suppress 50Hz
I suggested Twin-T Notch Filter
R1 105k
R2 105k
R3 31k
C1 47n
C2 33n
C3 47n

On paper it looks great, but how will it / won't work with real components?
C SMD NPO normal accuracy
R SMD 1% composed of two resistors for example R1 measures the actual resistance of the 100k resistor and and I will add a second resistor from the E24 series so that the result is as close as possible to 105k

Filter is connected directly to mains voltage via 2pcs 10nF capacitors with voltage resistance 2kV each.

 

[danadakk]

If you do the sensitivity analysis of a Twin T that will give you an idea
of tolerance versus worst case suppression. Or pspice sim and do
monte carlo and/or multistep sym.

This is for high speed but has interesting architecture -

https://www.ti.com/lit/an/slyt235/slyt235.pdf?ts=1640266189597

Some papers -

A general approach to twin-t design and its application to hybrid integrated linear active networks | Semantic Scholar

This paper approaches twin-T design with a view to controlling the sensitivity of the transmission zero with respect to component variations, according to criteria that are of particular interest in the design of hybrid integrated linear active networks. In this paper we approach twin-T design...

www.semanticscholar.org

https://bib.irb.hr/datoteka/203790.A3L-K3.pdf

Regards, Dana.

 

[crutschow]

The Twin-T passive 50Hz notch filter is down nearly 4dB at 180Hz.
Is that acceptable for your requirements?

If not, you may need to go to an active notch filter with a higher Q (sharper notch).

 

[carpenter]

If I am specific, the mains voltage is given by Un 230V 50Hz. The tonal voltage that I need to drive is between 0.2 - 1.2% Un, frequency 170-1600Hz, actually several selected frequencies 180 and higher.

The first attempts at pure software detection looked like this:
1VA transformer 230 / 6.3V, 100nF isolating capacitor + resistive divider voltage 1.65V + - 1V converted to ADC sampling at 1kHz. Using the Goertzel algorithm, the presence of a tone signal at the selected frequency is detected.
It works, but I'm afraid there will be problems with the extremely low level of the tonal signal, so I want to strengthen the weight of the tonal signal. Active filter is a nice thing, but relatively expensive and kills the magic of software solutions.
That's why I proposed a passive solution.
Large expensive transformer will be replaced by 2 pcs SMD C NPO 10nF 2000V, followed by the mentioned Twin-T Notch Filter and then ADC

According to the simulations (ideal component values), the attenuation is at 50Hz -83dB at 180Hz approx. -27dB ,in other words, the weight of the tone signal at the input to the ADC increases by 56dB .

How to compare?
The extreme case of tonal voltage is 0.2% of the mains . Ut=460mV Un=230V
In the first case, the transformer has a conversion of 1:36 to "ADC" is Ut = 13mV and Un=6,3V , ratio 1:485.
In the second case,the Ut = 22mV and Un=0,36V, ratio 1:16.

At first glance, this should work great.
But I have no practical experience with this and I am usually very careful in anticipation of very strong filters.

 

[FvM]

Your control receiver should be prepared for non-ideal mains voltage, up to 8% THD and 5 % telecontrol signals must be expected according to power quality standards, e.g. EN 50160. In so far, more than 20 or 30 dB suppression of mains fundamental doesn't increase the worst case signal to noise level respectively reduce the required ADC input voltage range.

--- Updated Dec 24, 2021 ---

Test levels for harmonic distortion and control signals are also specified in EN61000-4-13.

 

[crutschow]

According to the simulations (ideal component values), the attenuation is at 50Hz -83dB at 180Hz approx. -27dB

Where did you get that?
The simulation below, from your referenced website, gives a notch of near -40db, and and attenuation of about -4dB at 180Hz.

 

[Audioguru]

I agree that typical 50Hz electricity has plenty of 100Hz and 150Hz distortion. Those frequency levels might also need to be reduced.