1
\$\begingroup\$

I am in need of an intermediate frequency (IF filter) for somewhere around 200 Khz as intermediate frequency. However, I know just the bare minimum of filter design ( learnt all that a couple of days back) and am relying on software tools to give me the circuit and component values.

Needless to say, my expectations are not met. This is the latest design which I have tried :

enter image description here

------->>

enter image description here

[ It is a 6th order active high pass filter ( Butterworth) . The op amps are cascaded, the third one being in the last image ; cutoff being 135 Khz. It would have been attached to a Low pass filter to complete the bandpass if everything was OK ]

From the various articles describing superhet receiver, I believe the IF filter has to be very sharp, else tuning is not proper. However, all the designs so far I have tried of active filters ( I actually started from 2nd order and kept increasing the order), I observe the following :

1) Even in this 6th order design, I am not able to get the required sharp cutoff.

2) The observed output (regular sine) for this high pass type starts to begin from around 200 Khz. There IS an output above 100 Khz, but with the shape of sine distorted. What I believe I need is AN ATTENUATED output, not A DISTORTED output( otherwise wont it create problem in demodulation stage ?), for signals lying outside the passband.

3) Lower order filters, obviously give a more poor performance.

Here are some outputs at different freq :

enter image description here

enter image description here

enter image description here

I also tried ( again using software tools) passive LC filters. However, they tend to heavily distort the signal even in the passband ( although they have very sharp cutoffs).

Now my question is :

1) Is the performance I am getting from this 6th order active filter adequate ( I am not interested in pro level performance; just adequate performance would do) for the purpose of IF filter ?

2) If not, then what is the solution ? (I dont think increasing orders would solve this)

Thank you.

\$\endgroup\$
  • \$\begingroup\$ Is this a simulation exercise or if not what opamps have you used. More detail of how the 6th order HP filter attenuates is needed. Maybe show a graph so it can be seen if they are performing as expected. \$\endgroup\$ – Andy aka Oct 1 '14 at 17:09
  • \$\begingroup\$ No. I have actually realized the circuits. LM358 is used (that's what is available in my region besides 741). Above 200 KHz, it passes fine. But I am more concerned about the stopband. \$\endgroup\$ – Plutonium smuggler Oct 1 '14 at 17:12
  • \$\begingroup\$ @Andyaka . I have added some outputs. \$\endgroup\$ – Plutonium smuggler Oct 1 '14 at 17:23
  • 1
    \$\begingroup\$ What is your bandwidth goal? Two common strategies for very narrow are to move the IF to 4 MHz or so and use a number of commodity microprocessor clock crystals. Or use an IQ design with an IF of zero and op-amp lowpass filters (the later especially if then sampling with an audio-class ADC for digital processing). You could also use 455 KHz and a resonator salvaged from an AM broadcast receiver. \$\endgroup\$ – Chris Stratton Oct 1 '14 at 17:42
  • 1
    \$\begingroup\$ You'd probably find it informative to dig up amateur radio designs (and online repostings thereof) from the 1980's and 1990's - ie, when designs were primarily transistors and linear ICs, but not yet software radio. I believe you can use the microprocessor crystal idea with either a transistor IF amp, or possibly that ever-popular TV IF amp chip. \$\endgroup\$ – Chris Stratton Oct 1 '14 at 17:53
4
\$\begingroup\$

The LM358 and LM741 do not have enough bandwidth to do what you want. At 200kHz the LM358 can only manage about 10dB of gain. It also has poor slew rate and a high phase shift at this frequency, resulting in bad filter performance and distorted output.

If you don't have access to better opamps then you will have to use 'passive' filtering. Conventional IF amplifiers use a tuned transformer between each stage, with a tapped primary and loosely coupled secondary to achieve high Q. The downside of high Q is narrow bandwidth, but by staggering the center frequency of each stage the overall bandwidth can be improved without compromising stop-band attenuation.

You could build your own transformers from scratch, but it might be easier to scavenge some from an old AM radio and modify them to suit (you might even be able to leave them on the PCB and use the entire IF circuit as is). Inside each transformer you should see a small capacitor in parallel with the primary winding, which creates the tuned circuit. Wiring another capacitor across this will lower the frequency. To get from 455kHz to 200kHz you need to add about 4 times more capacitance (might be 2~10nF, depending on the value of the original capacitor).

The circuit below shows the IF part of a typical transistor radio. I have highlighted the capacitors that are in each IF transformer. The transformers are often coded with a color to indicate what stage they are in:- yellow = 1st, white = 2nd, black = third.

enter image description here

\$\endgroup\$
1
\$\begingroup\$

The LM358 IS the culprit - at 100kHz it has a miserable open loop gain of about 20dB and will be responsible for the distortion you see in the frequencies at around 135 kHz. At 200kHz, the filter is appearing to let stuff through adequately because the op-amp has virtually given up doing what it should be doing at this frequency and any signal you see is due to input-output coupling thru C1 and R2.

Look at the tool you used - it clearly says the minimum op-amp GBWP needs to be nearly 10MHz on one circuit and 26MHz on another: -

enter image description here

The LM358 has a gain-bandwidth-product of 1MHz - it is not suitable.

\$\endgroup\$
  • \$\begingroup\$ I dont understand; GBP = Gain * Bandwidth. So here, my gain = 1, shouldn't the op amp work till 1 MHz then ? Where's the flaw ? \$\endgroup\$ – Plutonium smuggler Oct 2 '14 at 2:54
  • \$\begingroup\$ If your op-amp circuit were just a unity gain amplifier then that is true what you say but it isn't - there is feedback via R2 and for correct operation of the filter you need the G*BW for the filter to work as expected. The tool you are using tells you this. The 358 is just not suitable for this application. Do you have a spice simulator to try a few different op-amps to give you a better idea? \$\endgroup\$ – Andy aka Oct 2 '14 at 7:14
  • \$\begingroup\$ I thought it's the negative feedback we take into account ( which is 1) while computing GBP. I do have simulators, unfortunately op amps other than 358 and 741 are not available. I am now trying to arrange / salavage crystals for crystal LC filter to get required performance, as suggested in comments above. \$\endgroup\$ – Plutonium smuggler Oct 2 '14 at 7:58

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.