How to design low-pass filter

[walalala]

Hi

I want to design a circuit to amplify low frequency and weak signal，so I need a LPF that corner frequency is 5kHz .I need to know which kind of filter is more suitable,active filter or LC filter? Is it right to choose Butterworth filter and then select the capacitance and resistance？ How to choose the order of filter , isn't it better to be higher？
Or is there something wrong my idea ? What should i consider on the design of filter?

Thanks

 

[stenzer]

Hi,

at low frequencies active filters are usually used, as components are getting pretty large at low frequencies. A butterworth filter is a good choice if you are interested in the excact amplitude of your signal, as this filter is maximum flat in the pass-band. To achive a high attenuation and a low transition between pass- and stop-band other filters like the elliptical one perform better (but not that flat in the pass-band).

A higher filter provides more attenuation but of course more components as well as space on your PCB is required, which causes also additional costs. So you have to identify your minimum required stop-band attenuation and transition-band width (transition from pass- to stop-band).

Have a look at the Filter Design Tool from Nuhertz [1]. You can download a 30 day free trial version. The software includes passive filters as well as active ones.

[1] **broken link removed**

BR

 

[KlausST]

Hi,

At first you need a clear definition what the filter should do regarding frequency and amplitude.
This means
* how flat it needs to be at frequencues below 5kHz
* how steep it needs to be above 5kHz.
"Flat" means how equal the gain at frequencies: DC, 10Hz, 30Hz, 100z, 300Hz, 1000Hz, 3000Hz (just example frequencies in your "pass band") is.
"Steep" means what gain (suppression = gain much smaller than 1) you need at frequencies 10kHz, 30kHz, 100kHz...(random frequencies in your "stop band")

For sure you may say you want perfect equal gain at pass band .... and perfect suppression in your stop band. But you need to see it from the other side: with how much "errors" or "imperfection" you can live. The closer you go to perfect the more difficult it becomes.

Active or non active. I agree that low frequencies call for active filters. 5kHz is low, especially for "L" when you need useful impedances in the low kHz range.
But a clear "must be active" is your requirement for amplification. As soon as you need a (flat) gain in your pass band greater than 1 you need an active filter.

A drawing will be good, a more detailed description what your filter is used for, too.

Klaus

 

[walalala]

Thanks for your advice.

Now i have decided to use active filter that gain is 1 , so there is a new question , how i select a suitable op amp as the filter or which property more important ? Precision , low noise , band , swing of input and output and others ? Or I select it straightway on the web of those companies like ADI , Ti .

What's more , r there some cheap but effective ways to check whether the op amp is genuine ? Last time i buy some LT1028 and the voltage supply trip while I connect it with pin .

Here is LT1028 datasheet

https://www.analog.com/media/en/technical-documentation/data-sheets/1028fd.pdf

The circuit page 17 shows was which i used before.

Thanks.

 

[KlausST]

Hi,

which property more important ? Precision , low noise , band , swing of input and output and others ?

I guess nobody can answer this for you. These are your or your application´s requirements.

LT1028: I used it, i like it. But it is not suitable for gain of 1. It is not stable for this gain.

And I think such an advanced OPAMP is overkill for your application. It is very low noise, designed for low (input) impedances.
Most applications don´t need this performance. I think it´s not easily available and expensive.

For a more detailed advice we need to know your requirements. ( But first you need to know )

Look for standard OPAMPs with unity gain stability (stable at gain of +1)

Klaus

 

[stenzer]

Hi,

I agree with @KlausST , the most important thing is to clarify your requirements. The Analog Filter Wizard [1] from Analog Devices may be of interest as it suggests suitable opamps according to your specifications.

[1] https://tools.analog.com/en/filterwizard/

BR

 

[walalala]

What you said above is useful . I have got it and can't wait to finish my circuit .

Thanks for helping me a lot .

 

[schmitt trigger]

Texas Instruments also has a very capable web based Filter Designer tool.

I haven’t checked lately, but Microchip had a similar tool.

 

[Audioguru]

Here is LT1028 datasheet

https://www.analog.com/media/en/technical-documentation/data-sheets/1028fd.pdf

The circuit page 17 shows was which i used before.

A "phono" or a "tape" preamp? I remember those old things.

 

[walalala]

A "phono" or a "tape" preamp? I remember those old things.

Right , I learned from some literatures that LT1028 could be LNA of nanovolt signal， so I want to use the reference circuit in the datasheet .

 

[KlausST]

Hi,

I'm confused.
Initially you talked about about building a filter.
Now it seems to be a extremely low noise application.

The LT1028 is low noise, but it has optimal noise performance on very low impedance signal sources.
Like phono pickups. You also see the 7.5 Ohms resistor in the feedback.

For higher impedances (input, feedback) there are better Opamps.
A filter circuit usually uses higher impedances. That what's worrying me.

I recommend to read through the Linear technology design note DN015: Noise calculations on Opamp circuits.

Klaus

 

[FvM]

In addition, an active filter for low level signals won't be used without an input preamplifier. This allows separate optimization, noise matching to the source impedance for the preamplifier, reasonable impedance level for the filter stage.

 

[crutschow]

A FvM noted, for lowest noise you would want to amplify the signal first with a low-noise amp to minimize the added noise from the active filter.
You could add a simple passive 1-pole low-pass filter at the input to the amp, if the noise is significant compared to the signal.