1
\$\begingroup\$

I am busy with a high order low pass filter. How can I improve the response time of a filter? If I use a higher ordered filter, does it improve? Also, if I use an op amp with a faster slew rate, does it improve?

My application is that amplify and get filter the load cell signal. My load cell signal is changing between 1 Hz to 10 Hz.

There is three stages in my signal conditioning circuit. First stage Instrumentation Amplifier to amplify the signal(AD8221), second stage is second order low pass filter with OP07C op-amp (Sallen-Key type, corner frequency is 10 Hz), third stage is voltage follower with again OP07C.

According to my test results, when I compared original signal to output of amplifier (first stage) there was no significant delay between two signals. However when I compared original signal to output of voltage follower there was significant delay. The delay was approximately quarter of a period of original signal.

So I have thought that I have to improve my filter rise time. Or do I have to change my op-amp with higher slew rate one. According to all your answer, should I use higher cutoff frequecy with same op-amp (OP07C)?

Most of you mentioned "group delay" of a filter. What does "group delay" depend on? I got it firstly depends on corner frequency of a filter. Is there any effect order of filter to group delay?

\$\endgroup\$
  • \$\begingroup\$ What is response time of a filter? \$\endgroup\$ – Eugene Sh. Jul 22 '15 at 14:04
  • 2
    \$\begingroup\$ To reduce the settling time, you need to increase the bandwidth. \$\endgroup\$ – Justin Jul 22 '15 at 14:17
  • \$\begingroup\$ Do you speak about step response (rise time) or about group delay? \$\endgroup\$ – LvW Jul 22 '15 at 14:57
  • \$\begingroup\$ is this an analogue filter or a digital filter? there are ways with digital filters and some NASTY ways (unrecomended) ways for analogue \$\endgroup\$ – JonRB Jul 22 '15 at 16:32
  • \$\begingroup\$ I want to reduce rise time of a filter. It is an analog low pass filter not digital. \$\endgroup\$ – Cem Jul 23 '15 at 7:04
6
\$\begingroup\$

You cannot greatly change response time of a low-pass filter by changing its' order. What you have to do is change its cutoff frequency - the higher the cutoff frequency, the faster the response.

Look at it this way. A low-pass filter removes high frequencies, right? And if you want the filter output to change more quickly it must contain more high-frequency components. You know, fast change means high frequency. So the only way to get faster response from the filter is to let more high-frequency signal through, and that means changing the cutoff frequency to a higher one.

\$\endgroup\$
3
\$\begingroup\$

Usually, the higher the order of the filter, the longer the response time so the trick is just filtering sufficiently to get a manageable signal-to-noise ratio. If the interferer is significantly away from the wanted frequencies, different filter types can improve response time of the wanted signal.

If you are looking to keep the time delay between two signals very close and only one of those signals needs to be filtered you can opt to filter both and therefore keep the time delays still acceptably close.

\$\endgroup\$
2
\$\begingroup\$

The behaviour of a filter circuit in the time domain and in the frequency domain is not independent on each other. Hence, you cannot change one without the other.

That means (as always in electronics): You have to accept a trade-off between the behaviour in both domains. Either you have specific frequency requirements (corner frequency, damping values) - and you have to accept the resulting time properties (group delay variations). Or you have certain requirements in the time domain (group delay, step response) and you have to live with the resulting frequency response.

The latter case is typical for Bessel-Thomson type filters which are selected primarily because of their good group delay properties (but have rather poor damping properties).

\$\endgroup\$
0
\$\begingroup\$

Generally speaking, unless extreme cases, slew rate in not important in filter design - unless the slew rate starts to interfere with the actual functioning of the design of the filter - distorting the signal for example.

Also, generally speaking, 'response time' increases with the order.

The selection of the filter architecture is the main factor in delay - delay is rarely constant with frequency, so you also have to decide if the filter has to show the same delay at each frequency or not. In a lowpass filter, delays tend to increase in the reject zone.

So, you need to decide the architecture... An interesting page on the subject is Analog Devices phase response comparison, where you can actually see the different responses as dependent on frequency.

If predictability of response time is a factor, you might want to consider migrating to digital filters. FIR filters have a constant time response.

\$\endgroup\$
0
\$\begingroup\$

I guess the cause of your delay is multiple phase-shift caused by using capacitors between amplifier cascades. For low frequency (10Hz you mentioned) the summary delay may be significant.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.