Continue to Site

Welcome to our site!

Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

  • Welcome to our site! Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

Active filters

Status
Not open for further replies.

confounded

New Member
Hi, i'm currently reading about active filters, but i'm pretty sure this is one subject in my textbook which is very outdated. Am i right in assuming that people just use a single IC whichever type of filter they require and add external components to achieve their requirement?
I have the MF10 universal dual switched capacitor filter datasheet and this IC will produce the classic filters in my textbook (butterworth, bessel, cauer and chebyshev) but it requires an external clock. Is there a wonderful IC that will produce any type of filter i want and requires minimun amount of external components, ideally no external clock?
What do you guys use as your filters?
 
ANY filter you say? ANY is a loaded word. There is no such IC because you can get first order filters, or filter of orders thousands of times larger.

Op-amps do not require oscillators though and can make many filters, but you may need a bunch of external components as well as more op-amps for higher order filters.

Switched capacitor filters will always require an oscillator of some kind, whether internal or external just like a microcontroller or microprocessor because it has a "sampling" nature like digital systems. This clock and sampling can interferre with other clocked systems if you do not use anti-aliasing filters or synchronize the clock with other clocks in the system.

A microcontroller/processor is a wonderful IC that can make any filter you want as long as it is has enough memory and is fast enough for your filter, and you don't need an external oscillator if the processor has an internal oscillator. But it outputs the filter in digital form so you need a DAC to convert it back to analog form (which no one ever does since we usually just use it in it'd digital form during processing). It also requires an analog anti-alias filter and ADC to convert analog data to digital before it can be filtered. This is called a digital filter and works by taking numerical samples and doing a bunch of math on them to come up with an output. So it is inherently slower than an analog filter, but you can get much much more ideal filter characteristics.
 
Last edited:
Continuous signal (non-switching) analog active filters are usually made with op amps and discrete components. Typically you can get a 2-pole filter per stage, but 3-poles are possible. Software such as Active Filter Design Application - FILTERPRO - TI Software Folder can make the design of such filters relatively easy.

There are some IC's, such as the UAF42 from TI, that are specifically designed for active filter design. The UAF42 includes some on-chip capacitors to minimiize the number of external components required.
 
thanks for replying guys.

I should of eplained what i was wondering existed better.
The classic filters i mentioned previously all have their own specific qualities and drawbacks eg butterworth has flat passband but poor steepness in transition region and poor phase characteristics, the bessel has good phase characteristics but large ripple in passband and poor steepness in transition region etc.
My textbook is quite old and so i am wondering how far technology has advanced.
Have they developed a filter that exibits such flat passband, fast transition from passband to stopband and constant time delay (good phase characteristics) that it is used whenever a filter is required and you simply add external components to produce either low pass, high pass, bandpass or whatever you wish and cascade this wonderful IC (and adjust additional external components) to produce the order height you want?
I am especially interrested in how you guys (pro's) make your filters, for non critical applications and more critical ones.
For example in audio applications, i have no idea but i assume the quality of filter is critical, what do you guys use there?
 
No such analog filter IC exists because the topology of the different kind of filters is too varied to combine all of them into a single IC. It also seems the ICs capable of more than one kind of filters need a LOT of external components to set everything up. Analog filter design is still labourous (and seriously out of my league).

Only digital FIR filters come close to what you speak of. They requires an ADC front-end followed by a processor to do the filter calculations. The more ideal the filter is, the more processor speed and memory is required. These filters can achieve perfect group delay. Throwing more calculations into the filter (aka more processor resources), you can approach a flat passband, narrow transition band, of arbitrary order, and very accurate frequency cutoffs for low pass, highpass, bandpass, and a bunch of other types of filters I've never heard of. It's all in the software. If the hardware is powerful enough, you just swap the software for a different filter. They are slower than analog filters though since they inherently have a processing latency (which is why you need a faster processor and ADC to process data at higher rates.

I suppose it's possible to add on a DAC to output the filtered signal in analog form, but no one ever does to my knowledge. Changing the information into digital form to be processed is usually the goal.
 
Last edited:
The design mathematics of filters were thoroughly analyzed many years ago and the math is still used today. The tradeoffs in filter types between cutoff sharpness, passband ripple, transient response, etc. are determined by the fundamental characteristics of the signals and circuits. There has not been (and is unlikely to be) any breakthrough that will lead to a magical universal filter.

Doing filters with digital processing may more easily allow you to generate higher order or more exotic filters, but they still all follow the same filter mathematics.
 
Thanks again for reply guys.

digital filters sound very interesting, i can't wait to get into the digital half of my textbook, so i can begin to understand them.

Also i downloaded the active filter design application FILTEPRO that crutschow linked, but after installing when i go to run it i get the following message:

Run- time error '339' ;
component 'comdlg32.ocx' or one of its dependencies not correctly registered: a file is missing or invalid.

Anyone know how to fix this? or at least whats wrong, i don't know much about computers. Any help would be much appreciated.
 
confounded, you know you can read the book when you're not in school =) The teachers only gonna repeat what's in the book.
 
Years ago I made excellent switched-capacitor Butterworth 4th-order and 8th-order filters with National Semi's ICs that had their own built-in oscillator. They are obsolete now but Maxim have similar ones.
 
Cool thing about parts like the MF10 is that you can make the filter CF adjustable by changing the clock frequency.
 
The MAX7401/7405 is an 8th order Bessel switched capacitor filter where the input clock frequency is 100x the cutoff frequency (between 1Hz and 5kHz cutoff frequency). It also has an internal oscillator that can be set by a capacitor. BUt it seems to be +/-25% accurate. I don't know about you, but isn't the reason that you use clocked filters is because you need a very accurate cut off?
 
Last edited:
I have never used the part and have not looked at the data sheet but, 25% sounds pretty awful, are you sure about that number?
 
THe internal oscillator (not using an external clock) seems to be 25% accurate from the datasheet. It is set with an external capacitor and it says that the Internal Oscillator Clock frequency when Cosc = 1000pF varies from 29-48kHz with 38kHz nominal.
 
If that is the case, I would say use external osc. :) Maybe some might not care so much about accuracy and opt for the cheaper option of internal.
 
Probably depends on operating temperature and supply voltage stability. If you used good caps and trimmed it properly you could probably get 1% or better if the voltage and temperature are stable.
 
Last edited:
wow, lots of replies, thanks guys.

I have no teachers, i'm self learning, you make a good point though, i probably need a newer textbook to compliment the one im using.

I'm still a little confused to how you guys make your filters when you need them? Do you go for IC filters like the MF10, design your own with op amps or use digital filters?

Surely a filter is such a common component of a circuit people have a prefered filter they use often and just 'chuck in' knowing its general properties?

ps still cant use that filterpro software :(
 
The MF10 is old and has poor specs. The LMF100 is the newer improved version.

There was the MF40 and another one then the LMF40 and another LMF were the improved versions.
I made many very low distortion sine-wave oscillators with the obsolete LMF40 4th-order Butterworth lowpass switched capacitor filter removing the harmonics. Some used the built-in oscillator that I trimmed with a trimpot to the exact frequency.
I made a notch filter with the LMF100 and some LMF40 lowpass filters as a tracking very low distortion analyser.

Some people feed a square-wave into a Butterworth switched capacitor filter and the output is a pretty good sine-wave.
I made a sinewave using a CD4017 in 10 steps and oversampled it with a switched capacitor lowpass filter for extremely low distortion.
 
Also i downloaded the active filter design application FILTEPRO that crutschow linked, but after installing when i go to run it i get the following message:

Run- time error '339' ;
component 'comdlg32.ocx' or one of its dependencies not correctly registered: a file is missing or invalid.

Anyone know how to fix this? or at least whats wrong, i don't know much about computers. Any help would be much appreciated.
It runs ok on my computer which is running Windows XP. What operating system are you using?

You might try reinstalling it.
 
Status
Not open for further replies.

Latest threads

New Articles From Microcontroller Tips

Back
Top