0
\$\begingroup\$

I designed a band-pass filter. I set the center frequency to 90 MHz. When I do an AC analysis, I see the filter graph I want, but the transient analysis results are not what I wanted.

I'm giving a single signal pulse to the input and I want to see that at the output, but the signal is distorted due to the filter.

I want to do good filtering without distorting the input signal.

What should I change in the circuit? What should I add?

I am sharing 2 filters that I have made with you.

The results of the filter circuit below are shown in the graphs. The frequency response of this filter is as I want it, but the results are very bad in the transient analysis. It distorts the input signal pulse.

Circuit:

enter image description here

Frequency response graph:

enter image description here

Filter input pulse:

enter image description here

Filter output pulse:

enter image description here

Now let's move on to the other filter circuit with a good transient result graph.

This filter circuit and its results are shown below.

Circuit:

enter image description here

Frequency response graph (this graph is not suitable.)

enter image description here

Filter input pulse:

enter image description here

The pulse I get at the comparator output (the results I want should be roughly like this.)

enter image description here

\$\endgroup\$
5
  • \$\begingroup\$ I'm signal is linear at best with a 30 dB range and the Bode plot has a ridiculous >400 dB range so if you compare the plots just looking at the top 30 dB and that higher Q is what makes the difference, not the DC \$\endgroup\$ Commented Jan 19, 2022 at 7:29
  • 3
    \$\begingroup\$ But Dc analysis results do not occur as I want. I'm giving a single signal pulse to the input That's not DC analysis. In a DC analysis, the signal does not change over time, so simulationg a pulse response is impossible in a DC analysis. What you're doint is a time-analysis called transient analysis. \$\endgroup\$ Commented Jan 19, 2022 at 7:50
  • \$\begingroup\$ What bandwidth do you want, and why? \$\endgroup\$ Commented Jan 19, 2022 at 9:30
  • 3
    \$\begingroup\$ You can't eat your cake and have your cake. If you want a specific transient response then design your filter to suit that. \$\endgroup\$
    – Andy aka
    Commented Jan 19, 2022 at 10:19
  • \$\begingroup\$ Are really needed R7, R14, R12, R16 ? In such filters ... are only needed the first and the last 50 Ohm. rf-tools.com/lc-filter \$\endgroup\$
    – Antonio51
    Commented Jan 19, 2022 at 13:04

3 Answers 3

2
\$\begingroup\$

The passive filter response looks normal for a highish-Q bandpass.

The active filter response is different because it doesn't work: the opamps don't have a negative power supply, yet the circuit would require them to output a negative voltage, so they're stuck in clipping when output voltage is near zero. In addition, the maximum input voltage spec of 4.5V is violated.

So you should use +/-2.5V power supplies for the opamps, and use a lower input amplitude.

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

Your analysis is far from being accurate. A BPF will slow down risetime ( due to -3dB BW) and cause ringing whenever Q>1. Its inverse is called Damping Factor. It seems you need a digital solution is a time delay, being a one-shot shorter than the input pulse.

To realize your transfer function, let's look at what you want.

+ve Pulse narrower

\$~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~\$Input: \$~~~~~~~~~~~~~~~~~~~~~~~~~~~~~\$Output

\$t_{PW50}\$, width \$~~~~~~~~~~~~~~~~~\$~some pulse like 40 ns\$~~~~~~~~\$< 10 ns
\$t_R\$ , rise time \$~~~~~~~~~~~~~~~~~\$ ~1 V/ns\$~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~\$same
V range\$~~~~~~~~~~~~~~~~~~~~~~~~~~~~~\$5 V\$~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~\$ same
\$t_{p}\$ Propagation delay: \$~~~~~~~~~~~~~~~~~\$ TBD
Load Capacitance [pF]: \$~~~~~~~~~~~~~~~\$ TBD assume 30 pF max

  • A typical trace over ground plane 3cm/ pF +IC load (let's assume 30 pF, a standard datasheet load).
  • A ground plane is needed to lower the impedance if prop. delay approaches or exceeds rise time.

Classic One shots are found in logic '123 package.

74HC123: Transition time 15 ns max @ 4.5V , PWmin = 45 ns typ @ 5V (TI) This does not meet your requirements.

74ALC123: is lower output capacitance (4 pF) and half the driver resistance. ~ 25 ohms typ @ 5V but 45 ohms max @ 4.5V = Vol/Iol

Pulse Width (min) = 5ns . This looks similar to your requirements which you need to get into a habit of articulating in a list such as above.

For faster risetimes, use Current Mode Logic (CML), which has fewer choices and excludes one-shots. But you can make one with NAND gates and an RC delay as I have done here.

Then you may achieve sub-nanosecond pulse widths with a suitable controller impedance board design with test points, tip/ring probes and/or SMC connectors, cable with 50 ohm load.

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

Made with microcap 12, spectrum-soft.com

Here an example (passive Butterworth standard) of a 90 MHz (center frequency filter), 10 MHz bandpass. Ok, there are some ... for any needs.

What one should get ... AC Analysis and TRAN Analysis (impulse).

One should note the "symmetry" of curves (AC and TRAN impulse).

enter image description here

enter image description here

enter image description here

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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