Digitally programmable high pass filter

[rfguy2020]

Hi,

I need digitally programmable high pass filter. Cutoff frequency should be adjustable 10kHz-150kHz. Bandwidth should be couple megahertz.

What is best way to design programmable high pass filter? I have been looking for digitally adjustable capacitors (DTC), but those are in pF level and I need something nF level.

[MasterT]

Digital potentiometer may be?  Microchip mcp4151 5 kOhm version has pass band up to 2 MHz

[David Hess]

What filter order and distortion level were you thinking of?

A low order low pass switched capacitor filter could be combined with a summing amplifier to generate a high pass response, and some switched capacitor filters are good to 200 kHz.

Operational transconductance amplifiers can be used to make current variable filter elements for any response that you want:  https://vahe.people.uic.edu/spring2016/ece412/OTA-structures2.pdf

Larger value capacitors can be switched into and out of a filter circuit using transistors or multiplexers.

[dobsonr741]

Switched capacitor single part universal analog filter block: https://www.anadigm.com/_doc/DS231000-U001.pdf

[radiolistener]

What is best way to design programmable high pass filter?

I think the best way to design digital filter is to use Matlab/Octave.
I tried different tools, but the only way that allows me to get the best result is playing with filter parameters in Matlab/Octave.
It has many instruments for filter design, testing and optimizations for your exact needs.

[Ground_Loop]

Maxim MAX260, 261, 262 family of digitally programmed filters may work for you. I’ve used them in audio range with PIC controllers.  The 262 should be good to 1.5 MHz.

[rfguy2020]

What filter order and distortion level were you thinking of?

A low order low pass switched capacitor filter could be combined with a summing amplifier to generate a high pass response, and some switched capacitor filters are good to 200 kHz.

Operational transconductance amplifiers can be used to make current variable filter elements for any response that you want:  https://vahe.people.uic.edu/spring2016/ece412/OTA-structures2.pdf

Larger value capacitors can be switched into and out of a filter circuit using transistors or multiplexers.

I have been looking for 1st order filter, but I'm not sure if I need higher order filter. Distortion level requirement is not clear yet.

I have tried to simulate large capacitors + multiplexer, but I have not been too successful. I'm not sure what kind of switch/multiplexer I need to use. I was thinking to put four different capacitors (1x, 2x, 4x and 8x) in parallel with ADG601 switches. Then I could get 16 different cutoff frequencies. For some reason, steps are not expected due to some parasitic effect?

[rfguy2020]

What is best way to design programmable high pass filter?

I think the best way to design digital filter is to use Matlab/Octave.
I tried different tools, but the only way that allows me to get the best result is playing with filter parameters in Matlab/Octave.
It has many instruments for filter design, testing and optimizations for your exact needs.

The challenge is that I cannot get my signal to ADC without analog filter. My signal has strong DC component and I need to filter that out before I can digitize the signal. The signal I need to digitize is not always the same and for that reason I need programmable analog filter.

[rfguy2020]

Switched capacitor single part universal analog filter block: https://www.anadigm.com/_doc/DS231000-U001.pdf

Interesting product and I have not heard about this before. It might be overkill for my project and it seems to be difficult to get chips from common sources.

[rfguy2020]

Maxim MAX260, 261, 262 family of digitally programmed filters may work for you. I’ve used them in audio range with PIC controllers.  The 262 should be good to 1.5 MHz.

I have been looking MAX26X as well. I tested LTC6602, which seems to be similar than MAX26X. I don't like switched capacitor filters, since the clock signal is causing noise. I would prefer to find continuous time filter solution.

[tszaboo]

What filter order and distortion level were you thinking of?

A low order low pass switched capacitor filter could be combined with a summing amplifier to generate a high pass response, and some switched capacitor filters are good to 200 kHz.

Operational transconductance amplifiers can be used to make current variable filter elements for any response that you want:  https://vahe.people.uic.edu/spring2016/ece412/OTA-structures2.pdf

Larger value capacitors can be switched into and out of a filter circuit using transistors or multiplexers.

I have been looking for 1st order filter, but I'm not sure if I need higher order filter. Distortion level requirement is not clear yet.

I have tried to simulate large capacitors + multiplexer, but I have not been too successful. I'm not sure what kind of switch/multiplexer I need to use. I was thinking to put four different capacitors (1x, 2x, 4x and 8x) in parallel with ADG601 switches. Then I could get 16 different cutoff frequencies. For some reason, steps are not expected due to some parasitic effect?

With analog switches, you want small capacitors and larger resistance values. Analog switch isn't exactly 0 Ohm, and it changes with temperature and voltage, so you want to minimize the susceptibility of the circuit on this error resistance.
IMHO it's a lot simpler to just prototype the circuit, and use the right value filter rather than switching circuits.
And if you already have an ADC in mind, place the lowest cutoff filter before it, and filter it digitally if necessary. Changing filter coefficients is easy, compared to verifying all that analog stuff.

[jonpaul]

no need for filter, use oversampling Delta Sigma ADC.

https://www.analog.com/en/resources/technical-articles/sigmadelta-adcs-tutorial.html
https://en.wikipedia.org/wiki/Delta-sigma_modulation
https://inst.eecs.berkeley.edu/~ee247/fa05/lectures/L24_f05.pdf


Sharp digital filter built in, Fcut is proportional  to sample rate.

Bon chance

Jon

[tszaboo]

no need for filter, use oversampling Delta Sigma ADC.

https://www.analog.com/en/resources/technical-articles/sigmadelta-adcs-tutorial.html
https://en.wikipedia.org/wiki/Delta-sigma_modulation
https://inst.eecs.berkeley.edu/~ee247/fa05/lectures/L24_f05.pdf


Sharp digital filter built in, Fcut is proportional  to sample rate.

Bon chance

Jon

OP wants a high pass filter. Delta sigma has a low pass filter.

[rfguy2020]

What filter order and distortion level were you thinking of?

A low order low pass switched capacitor filter could be combined with a summing amplifier to generate a high pass response, and some switched capacitor filters are good to 200 kHz.

Operational transconductance amplifiers can be used to make current variable filter elements for any response that you want:  https://vahe.people.uic.edu/spring2016/ece412/OTA-structures2.pdf

Larger value capacitors can be switched into and out of a filter circuit using transistors or multiplexers.

I have been looking for 1st order filter, but I'm not sure if I need higher order filter. Distortion level requirement is not clear yet.

I have tried to simulate large capacitors + multiplexer, but I have not been too successful. I'm not sure what kind of switch/multiplexer I need to use. I was thinking to put four different capacitors (1x, 2x, 4x and 8x) in parallel with ADG601 switches. Then I could get 16 different cutoff frequencies. For some reason, steps are not expected due to some parasitic effect?

With analog switches, you want small capacitors and larger resistance values. Analog switch isn't exactly 0 Ohm, and it changes with temperature and voltage, so you want to minimize the susceptibility of the circuit on this error resistance.
IMHO it's a lot simpler to just prototype the circuit, and use the right value filter rather than switching circuits.
And if you already have an ADC in mind, place the lowest cutoff filter before it, and filter it digitally if necessary. Changing filter coefficients is easy, compared to verifying all that analog stuff.

It seems that ADG601 has very low resistance, which should minimize loss. I went for ADG612, which has higher resistance but it seems to work better in simulator. I'm not sure if the difference is caused by charge injection or Cs/Cd... Or something else.

[kripton2035]

pretty sure you could achieve this filter by programming a cypres psoc.

[radiolistener]

The challenge is that I cannot get my signal to ADC without analog filter. My signal has strong DC component and I need to filter that out before I can digitize the signal. The signal I need to digitize is not always the same and for that reason I need programmable analog filter.

you can remove DC with a simple capacitor, non flat response can be compensated with FIR filter in digital domain.

[rfguy2020]

The challenge is that I cannot get my signal to ADC without analog filter. My signal has strong DC component and I need to filter that out before I can digitize the signal. The signal I need to digitize is not always the same and for that reason I need programmable analog filter.

you can remove DC with a simple capacitor, non flat response can be compensated with FIR filter in digital domain.

It is not that simple, since my DC level is not constant (so it is not actually DC). That is the reason I need programmable HPF, which I can adapt for changing "DC" level.

[RFDx]

I have tried to simulate large capacitors + multiplexer, but I have not been too successful. I'm not sure what kind of switch/multiplexer I need to use. I was thinking to put four different capacitors (1x, 2x, 4x and 8x) in parallel with ADG601 switches. Then I could get 16 different cutoff frequencies. For some reason, steps are not expected due to some parasitic effect?

Switch resistors instead of capacitors. See attached example, a single supply, adjustable 2nd order Sallen & Key HP. The filter has ~3dB gain in order to have same value resistors in the filter sections. With bipolar supply there is no need for the 2nd opamp (U2) and associated components.