Low pass filter failing at the lower range

[atferrari]

Please let me ask about this without showing any circuit, YET.

There is an active 4-poles Sallen Key lowpass filter. Cut off freq (-3 db) = 740 Hz. Both LM 358 opamps fed with +/- 9 V and duly filtered.

At the input, the signal from a network of 12 resistors driven by CMOS shif registers generating sine sequence of 24 voltage steps. Their output is referred to ground and so input to filter properly centered on 0V.

To my surprise, from 120 Hz to 500 Hz (upper limit of this design), filter output is remarkably smooth and looks OK, but going down from 120 to 10 Hz, the filter output is more and more jagged (one peak for every step at the input).

My question= is it any typical reason for this filter to behave like this? What would you check first in a case like this?

Just in case, please note that I tried two designs already: one with equal value resistors and the current one designed with Filterlab from Microchip.

I am far from my bench so posting values/circuits is not feasible by now.

My intention was to generate those steps in LT Spice to see what could be the reason but (see my post somewhere below) could not find how yet.

Do lowpass filters behave like this?

Agustín Tomás

 

[sreid]

Are you changing the input signal frequency by changing the shift register shift frequency? If so, the lower in frequency you go, the more low frequency components the signal will have.

 

[IRstuff]

Sounds like you have series capacitors in the path. The peaks would seem to represent the step transitions. As suggested by the SREID, this may be a consequence of reducing the output step frequency.

TTFN

FAQ731-376

 

[logbook]

>Do lowpass filters behave like this?
YES!

Your circuit is working as designed. You have a low pass filter at 740Hz. It will not filter signals at 500Hz and below. Why does it work for your 120Hz signal? Well, you are reconstructing the 120Hz. This requires higher frequency components such as double, treble and other integer multiples of the fundamental frequency. Google "harmonics".

As you go lower in frequency you get less and less filtering of the harmonics. The steps in the waveform are high frequency signals (harmonics).

 

[Comcokid]

I think sried and others have pointed out the issue. At the low frequencies, the switching harmonics are still within the bandpass of the filter.

You could try scaling down the filter when generating lower frequencies. I have seen this done by using SPST IC switches to add-to or double the capacitor values to cut in half the filter cutoff frequency.

Also, if you are varying some clock to change your signal, then you might look at using a switch-cap filter IC for your filter. By varying the clock to a switch-cap filter IC, you vary the cutoff frequency. If you scale your clock correctly between your signal generating circuit and your switch cap filter, then the filter will track the generated signal.

 

[IRstuff]

I thought the OP was complaining that the low frequency response of his low-pass filter wasn't working correctly. A low-pass filter should show unity gain at 0 frequency.

Perhaps the OP can submit a schematic or point to one on the web. Perhaps your filter is actually a bandpass filter, and not a low-pass filter

TTFN

FAQ731-376

 

[Noway2]

After re-reading the post a few times, it looks to me as though the issue is that the filter isn't behaving like the OP wanted, but that it is working as designed.

He says that he is applying a step wave form to a low pass filter with a 740Hz cutoff freq. He says that with inputs with a fundamental between 120 and 500 Hz that the output looks good. I take this to mean that he is seeing a relatively smooth sine wave of the fundamental frequency. As was already pointed out, this would be expected since at least most of the harmonics (depending on input freq) are being significantly attenuated. At low input frequencies, 10Hz-120Hz the output looks choppy which of course it would sine the 3rd and 5th and likely even the 7th harmonic are clearly in the pass band.

 

[IRstuff]

The OP states:

jagged (one peak for every step at the input).

for the lower frequency.

Presumably, the OP would still recognize his blocky sine wave. A "peak" would imply a differentiation of the steps in the waveform, implying a highpass behavior at the lowest frequencies.

TTFN

FAQ731-376

 

[jimkirk]

Sallen-Key describes the topology of the circuit. the OP says little about the response other than it's low pass with 4 poles and -3 dB at 740 Hz. It's not necessarily a Butterworth or Bessel. And even a 4 pole Butterworth will exhibit some overshoot to a step input. Get the Q wrong and who knows what could happen.

Get a network analyzer (or even a swept sine source and a scope) and measure the response.

 

[atferrari]

The first file is the circuit and the other four are pictures taken when output frequency was appr 77 Hz.

Their names are self explanatory, I hope.

http://img59.imageshack.us/img59/678...circuitpk2.png

http://img261.imageshack.us/img261/7...networkxd7.jpg

http://img403.imageshack.us/img403/9...networkrf7.jpg

http://img527.imageshack.us/img527/2...eforercin1.jpg

http://img257.imageshack.us/img257/6110/afterrcrg5.jpg

I can see that I would end redesigning everything. I prefer not to increase the number of steps. Do I have other chances even if using a more complex filter?


Agustín Tomás

 

[IRstuff]

Only one scope picture seems to be in existence.

TTFN

FAQ731-376

 

[atferrari]

Yes, links do not work.


My circuit

http://img374.imageshack.us/img374/5610/lowpasscircuitdk1.png

Network not loaded by filter

http://img514.imageshack.us/img514/364/unloadednetworkvv4.jpg

Network loaded by filter

http://img404.imageshack.us/img404/4933/loadednetworkob0.jpg

After filter, before RC passive highpass

http://img412.imageshack.us/img412/2176/afterlowpassbeforercvo2.jpg

After RC passive

http://img413.imageshack.us/img413/3928/afterrcdv2.jpg

Agustín Tomás

 

[IRstuff]

Your C5 makes it a bandpass filter, automatically. There is no DC response, hence, you do not have a low-pass filter.

TTFN

FAQ731-376

 

[jimkirk]

What are the values of R1 through 12? They, along with R21 make up a source resistance in series with R13, and that will affect the response of the first stage.

The 2 uF along with 5.4 kOhms (27k in parallel with 6.8k) rolls off below about 15 Hz, which is is pretty low, probably showing up in the phase shift between the fundamental frequency and the ripple in your last two pictures.

 

[LionelHutz]

The problem is starting right at your source. I'm guessing the values of R1 to R12 are large compared to R13 and that isn't working. As you change the steps you effectively change the size of R13 which messes up the filter. Try a unity buffer stage before the filter.

Even then, the comments about higher harmonics passing through the filter at the lower frequencies probably still apply and you will never fully get rid of the level steps with this approach.

R18+R19 are effectively in parallel with R17 making the frequency of that circuit much lower than you probably expected too.

Basically,
The filter load is changing the nice "steps" of your source to those peaked shapes.
The high roll off frequency of the filter is passing the distorted steps to the output.
The highpass frequency of the DC decoupling? capacitor/resistor circuit is then distorting the signal a bit more.

What is the purpose of the circuit? There are likely easier ways to generate a sinewave such as using a higher frequency PWM that is easy to filter. Or, find a function generator IC that does what you want.