I would like ot design a band pass active filter. Pass band is 20Hz to
600kHz. The required gain is 1200.
I am planning to use first order High Pass Filter with cutoff frequency 20Hz, gain 30, cascaded to second order Low Pass Filter with cutoff frequency 600kHz.
Using Texas Instruments' Filter Pro software, I could design a Low pass filter and high pass filter with Real Pole, Chebyshev configuration.
I am planning to use TI OPA4277.
I don't have any experience before in testing and designing such filters.
My wonder if I might need any additional capacitor, industor for impedance matching while cascading the high pass circuit to low pass circuit.
Please let me request for your suggestions and comments on it.
Active filters have their signal at the output of opamps, which are low impedance, thus just cascade the filters. The stages do not interact.
Usually you want your gain at the front end, so I'm not sure why you picked a gain of 30. Now there is a point where you run out of GBWP in the op amp in the highpass. That is, too much gain there taxes the opamp. It is something to study. But I would put as much gain the in the front as possible. One idea would be a HFP LPF HPF cascade if you can use two poles of HP filtering. That scheme would leave you with the offset of just one opamp at the output stage, plus roll off some of the 1/f noise of the LPF. When cascading filters, you usually cascade them in the order of increasing Q. When I did SCF, that gave the best noise performance.
When designing active filters, you need to realize that the output of every op amp is a filter output, just not the filter you intend to use. However, what you need to do is sweep the input of the cascaded filter over frequency, and note where each op amp peaks. Depending on topology, you can adjust the op amps so that they all peak at the same level, though not necessarily and most likely not the same frequency. This is known as dynamic range adjustment and often overlooked by first timers doing active filters.
I did a quick google search and didn't find anything useful. OK, I found a google books on line telling you to dynamic range adjust, but didn't spell out the procedure. I generally build ladder filters, so it is part of the game. I have no idea what the TI software does.
The idea is to run spice and plot the output of every opamp as you sweep frequency. You need to adjust the elements associated with the op amp in a manner where you only adjust the gain but not shift the poles. Since ladder filters are always prototyped with a signal flow graph, you can look at the do this visually. As you amplify one node, you make to reduce the gain in the node that use the signal from that opamp. This is easy to see in a signal flow graph since you amplify the paths into the integrator and attenuate the paths that look at that integrator.
If you have a node that doesn't achieve the peak signal swing, it just makes the filter a bit noisier. The real problem is when there is gain at a node that exceeds the rest of the filter. Then you will have that op amp clip, and in doing so determine the maximum swing at the output.
I second sourced a LTC part where their designer failed to dynamic range adjust. Actually their whole design was a mess in other respects. It took quite a bit of work to convince marketing to make a second source that was electrically different from the primary source, but the signal swing was a problem in the primary source, so marketing agreed with the change.
The good news is there will be no peaking with the single pole HPF. The LPF should be operating where there is little effect from the first HPF, so probably you could look at the LPF just by itself regarding dynamic range adjustment. This won't always be the case in general, but for the single pole HPFs, it is true.
If you expect the input signal to have sharp edges and the first stage is a high-pass its a good idea to have a single low-pass RC at the input before the first op-amp. This can prevent slew-rate limiting which might otherwise cause distortion and give an unexpected frequency response.
This isn't hypothetical - I was once caught out by this. A high-pass filtered pulse train intended for psycho-acoustic testing had a much greater low-frequency output than had been intended until the rise time of the pulse was tamed a little.
Hmm that's a nice 'DC' opamp. 10uV of VOS. But you don't need that in this application. And with a GBW of only 1 MHz, you are going to have a hard time getting a gain of 1200 at 600kHz. You should find a faster opamp. If your signal swings are large at the output you will also care about the slew rate too.
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.