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 Filter Design

Status
Not open for further replies.

boozi

New Member
Hello guys,

I'm currently a student at the University of Maryland. I would really appreciate it if someone could help me out with the following:

I'm currently doing research with some other students on active filter design (for a vibrometer). We are trying to design an active bandpass filter with a center frequency of 80MHz and a 3dB bandwidth of about 1MHz (Q factor of 80). We tried the multiple feedback and biquad bandpass ones. Now, everything worked completely fine when we simulated the designs on PSPice (Orcad) with the ideal opamp. However, when we replaced the ideal opamp with commercial ones (for example, TI's THS4271), the results we got were very, very far off.

So! My question is.... When I want to look for a suitable opamp, what specifications should I mainly focus on? The GBW should be more than 80MHz, correct? I'm not exactly sure what I should look for if I want the 3dB bandwidth to be 1MHz. Any suggestions you have for me? We've been trying quite a lot of opamps, but none would give us the results we want unless we change the values of our resistors and/or capacitors.

Thanks!
 
Welcome to the real world. Designing active filters at 80MHz is not trivial. You need a high bandwidth amplifier, careful pcb layout with good ground planes, good decoupling of all amplifiers and very short connections to the amp summing node (with no ground plane under the summing node to minimize node capacitance). At that frequency it's very easy to have your amplifier become an oscillator.

Depending upon the design, you probably need an amp with a GBW of likely ten times or more the highest frequency of interest (or 800MHz in this case). That's a very high frequency op amp. You can help determine that by simulating your filter with various GBW op amps. The ideal model may allow you to vary that parameter to see this effect. After you determine what you need, then you can select the appropriate amp.

You might consider a Sallen-Key topology which only requires a gain-of-1 follower amp. Those may be easier to obtain with high frequency characteristics then a standard op amp. You can even use a single-transistor, emitter or source follower for this amp.
 
Thanks for the reply crutschow. We already did what you said. The opamps that we tried to use had a GBW of 1GHz or higher (THS4271 -> 1.4GHz, etc). The funny thing about it, though, was that when we tried it with our parameters to achieve the desirable results (fc = 80MHz, 3dB bandwidth = 1MHz, and a gain of 1), the center frequency dropped all the way down to something between 8-12MHz. We tried playing around with our passive parameters, but it pretty much gave us the same thing. However, when I set the quality factor to 1 (as opposed to 80), the center frequency went all the way up to 67MHz (but then again, the 3dB bandwidth was extremely high -> 80MHz). Any thoughts on that?

In any case, I'll try and do what you suggested. Hopefully, it'll work out.
 
If you post your circuit, I'll take a look.
 
I'll try and do it tomorrow as I do not have PSPice on my computer. Thanks again for your help!
 
So here it is. I also provided the bode plot I got with matlab. The simulation on PSPice gave the same results. However, when I switch the ideal opamp with THS4271, the stuff I described before happens. I've also attached THS4271's data sheet. Let me know what you think.
 

Attachments

  • MFB Filter.jpg
    MFB Filter.jpg
    30 KB · Views: 367
  • Bode Plot.jpg
    Bode Plot.jpg
    35.8 KB · Views: 419
  • ths4271.pdf
    1.1 MB · Views: 230
You need much smaller value resistors. The input capacitance of the device will roll off your frequency. In the data sheet they are showing circuits with resistors below 1000 ohms.
 
If that's the case, wouldn't R2 be extremely small? In order to get the wanted results, if I make R1 & R3, R2 will have to be really small.....
 
You can simulate this all month long, but you will never realize it as an actual physical circuit. Think about this: each pin of an IC package has an intrinsic parasitic capacitance of several pF. Pin to neighboring pin capacitance is ~0.5pF, the stray inductance of pins and PC traces also becomes significant at these frequencies...
 
So in order to avoid parasitic capacitance, what we thought of doing is actually replacing the op amp with a transistor, and if the circuit is "small" enough, solder all the parts together without a board (it'll be a big mess, but we just want to see if it'll work).
 
You can reduce parasitic capacitance, not avoid it. And getting it below 1pF is essentially impossible.

Transistors also have parasitic capacitance so that won't necessarily help you.

The real world can be contrary.
 
Go with LC tuned circuit(s). The parasitics are less problematic; they become part of the tuning network.
Do you have access to a RF Network Analyzer?
A Sweep Generator?
A RF Signal Generator?
A 400MHz O'Scope?
 
I suggest that the bandpass filter circuit you are trying to use is not practical when scaled to these frequencies.

If you wish to continue using active filters, try cascading low-pass and high-pass filters to give a bandpass response.
Have a look at The Art of Electronics by Horowitz and Hill, Chapter 5 Active filters and oscillators, (my copy is the Second Edition, other editions may vary).

Otherwise MikeML's suggestion of using passive LC filters is a good one.

JimB
 
Much appreciated guys. I will take a look at that book. I'll let you know how it goes. By the way, we already tried an LC filter last semester, but that didn't work out as well as we wanted. Anyways, thanks again! I'll post an update later this week.
 
... By the way, we already tried an LC filter last semester, but that didn't work out as well as we wanted...

Probably because you didn't have (or know how to use) a RF Network Analyzer,
A Sweep Generator,
A RF Signal Generator,
A 400MHz O'Scope.
 
We tried it on a spectrum analyzer, but "funny" stuff happened....
In the end, though, I changed a few components, and it somewhat
worked... Not as well as we wanted, though.

Btw, JimB -> Do you think that the THS4271 will work with the
cascaded filter you suggested? Well, what I mean by that is,
will it be possible to achieve the wanted results with that op-amp?
 
Last edited:
Btw, JimB -> Do you think that the THS4271 will work with the
cascaded filter you suggested? Well, what I mean by that is,
will it be possible to achieve the wanted results with that op-amp?

I dont see why not.
You will have to build the circuit using VHF RF techniques, but that should not be too big a problem.

JimB
 
So I've tried two circuits that were in the book you suggested (well, one is a "special" case of the other one). I calculated everything and here are the results. As you can see, all is well with the ideal op amp.

I'm pretty sure I'm doing something wrong with the real op amp. Technically, the commercial one is supposed to be as close as possible to the ideal one, but since it's not, the equations I used to calculate the passive elements (Rf, Rq, Rg) would not really work, right? Sorry for asking so many questions. I just want to understand as much as I can about the design.

P.S. My apologies for the messy circuits.
 

Attachments

  • Schematic of Ideal.JPG
    Schematic of Ideal.JPG
    51.4 KB · Views: 222
  • Schematic of THS4271.JPG
    Schematic of THS4271.JPG
    75.1 KB · Views: 208
  • Simulation of Ideal.pdf
    12.5 KB · Views: 178
  • Simulation of THS4271.pdf
    17.7 KB · Views: 174
Last edited:
I was thinking more of the LP and HP filters in figure 5.16, rather than the tour de force of figure 5.18.

(All references are Horowitz and Hill the Art of Electronics)

JimB
 
Status
Not open for further replies.

Latest threads

New Articles From Microcontroller Tips

Back
Top