Even though we publish new blogs daily from your intrepid editor and a number of well-known writers, sometimes it's good to have a look back at our well-received articles from the recent past.
We touched on the basics of active filter designs once before when we looked at integrators in Op-Amp & Integrator Basics. This time, we go in deeper and look at active filters with more than one pole. Like the integrator, the active filter is constructed around an op-amp. The author of this article looks at the two common topologies: the Sallen-Key (a non-inverting filter); and the infinite-gain, multiple-feedback (an inverting filter):
The author also discusses sensitivity issues — how the filter behaves as the resistor and capacitor values deviate from the calculated values. All good stuff to know for your next filter design. Have a look here.
9 comments on “Active Filters: A Signal Chain Basics Review”
eafpres
April 8, 2013
Hi Brad–you mention in the linked article that usually above a few MHz the filters are done with passives. I've seen plenty of that done, along with matching networks in RF systems. I was wondering what the highest frequency application you have seen using and op-amp based active filter is?
Also, we have talked elsewhere about active front ends for RF systems; these circuits seem like the original genesis of those devices. This is a bit off this topic but the holy grail in, say, a mobile phone would be an integrated active match for the antenna plus active filter, ideally able to switch among different bands and have very low insertion loss in-band. What limits come into play to prevent designing the perfect solution?
I would say you can easily build op-amp based filters at 100MHz, for just 2nd order low-pass even higher. For RF matching and low noise, inductors are needed! On-chip coils have bad Q, but with active LC filters you can handle even 2GHz and built calibrated narrow-band filters. Howver, the noise of that is not as good as it would be for a LC filter with high-Q elements. But it could be good enough for a post-LNA pre-Mixer RF filter, and not ultimate-performance RX systems.
I would expect at 100MHz the parasitics would start to be problematic – stray capacitance and inductance. And as you've noted, when you actually want an inductor, it's tough to make one that's hi-Q. But I'll say up front that this is not my area of expertise.
Both the architectures are great for implementing a 2nd order filters compared to other architectures like Tow Thomas Biquad…
The design equations for MFB are straight forward with a few assumptions while Sallen Key has few more set of equations to solve, which means the inter dependency is more… I feel MFB is better when we consider variation or sensitivity of Ao or Q of the circuit w.r.t. to the passive elements.
eafpres: I have seen these kind of second order filter implementations using opamps and passive networks using discrete components… But are these architectures used for on-chip implementations?
I feel many designers either use the RC filters or gm-C filters for on-chip implementations… Any details ??
If we go as per design equations then Sallen key might be stable architecture, but the positive loop is a concern, we have to ensure the negative feedback is always more than the positive feedback.
We dont have that kind of concern is MFB circuits, but MFB circuits have 2 loops. Deciding which loop is vulnerable, which loop to break for stability analysis is complex.
If you use some of the look-up tables for calculating values, you sould have a circuit that is stable (i.e., properly damped). If you design from scratch, certainly care must be taken – again, got to keep damping where it belongs/keep Q from getting to high. Otherwise, you end up with an oscillator.
“sometimes it's good to have a look back at our well-received articles from the recent past” — This is why I personally enjoy printing out some of my favourite articles. I'm not always meticulous about doing it because I'm pretty sure that there are a lot of articles out there that I just haven't gotten around to reading. But for the ones that have been helpful, there's folder of those somewhere for reference in my study!
You must verify your email address before signing in. Check your email for your verification email, or enter your email address in the form below to resend the email.
Please confirm the information below before signing in.
{* #socialRegistrationForm *}
{* firstName *}
{* lastName *}
{* displayName *}
{* emailAddress *}
By clicking "Sign In", you confirm that you accept our terms of service and have read and understand privacy policy.
{* /socialRegistrationForm *}
Registration
Please confirm the information below before signing in. Already have an account? Sign In.
{* siteName *}
By clicking "Create Account", you confirm that you accept our terms of service and have read and understand privacy policy.
{* /registrationForm *}
Thank You for Registering
We have sent a confirmation email to {* emailAddressData *}. Please check your email and click on the link to verify your email address.
Hi Brad–you mention in the linked article that usually above a few MHz the filters are done with passives. I've seen plenty of that done, along with matching networks in RF systems. I was wondering what the highest frequency application you have seen using and op-amp based active filter is?
Also, we have talked elsewhere about active front ends for RF systems; these circuits seem like the original genesis of those devices. This is a bit off this topic but the holy grail in, say, a mobile phone would be an integrated active match for the antenna plus active filter, ideally able to switch among different bands and have very low insertion loss in-band. What limits come into play to prevent designing the perfect solution?
I would say you can easily build op-amp based filters at 100MHz, for just 2nd order low-pass even higher. For RF matching and low noise, inductors are needed! On-chip coils have bad Q, but with active LC filters you can handle even 2GHz and built calibrated narrow-band filters. Howver, the noise of that is not as good as it would be for a LC filter with high-Q elements. But it could be good enough for a post-LNA pre-Mixer RF filter, and not ultimate-performance RX systems.
Bye Stephan
I would expect at 100MHz the parasitics would start to be problematic – stray capacitance and inductance. And as you've noted, when you actually want an inductor, it's tough to make one that's hi-Q. But I'll say up front that this is not my area of expertise.
Both the architectures are great for implementing a 2nd order filters compared to other architectures like Tow Thomas Biquad…
The design equations for MFB are straight forward with a few assumptions while Sallen Key has few more set of equations to solve, which means the inter dependency is more… I feel MFB is better when we consider variation or sensitivity of Ao or Q of the circuit w.r.t. to the passive elements.
And we know that some versions of the active filter show sensitivity to minor component variations (per the author).
eafpres: I have seen these kind of second order filter implementations using opamps and passive networks using discrete components… But are these architectures used for on-chip implementations?
I feel many designers either use the RC filters or gm-C filters for on-chip implementations… Any details ??
If we go as per design equations then Sallen key might be stable architecture, but the positive loop is a concern, we have to ensure the negative feedback is always more than the positive feedback.
We dont have that kind of concern is MFB circuits, but MFB circuits have 2 loops. Deciding which loop is vulnerable, which loop to break for stability analysis is complex.
If you use some of the look-up tables for calculating values, you sould have a circuit that is stable (i.e., properly damped). If you design from scratch, certainly care must be taken – again, got to keep damping where it belongs/keep Q from getting to high. Otherwise, you end up with an oscillator.
“sometimes it's good to have a look back at our well-received articles from the recent past” — This is why I personally enjoy printing out some of my favourite articles. I'm not always meticulous about doing it because I'm pretty sure that there are a lot of articles out there that I just haven't gotten around to reading. But for the ones that have been helpful, there's folder of those somewhere for reference in my study!