Homebrew CW audio filter for uBITX - thumping/distortion on leading edges


Rob French (KC4UPR)
 

Hello,

I've implemented a simple audio band pass filter for CW on my uBITX v5.  It uses 3 sections of a TL974 op amp running off of 5V regulated (the 4th section of the op amp is used a buffer for a line out).

The three sections are identical, and implement a bandpass filter with a center frequency of about 700 Hz (I've attached a screenshot of the audio response as I tuned across a CW signal on 40 meters).

The problem I have is that when a signal is tuned up in the passband, plus or minus a couple hundred Hz, I get a strong thumping or distortion effect on the leading edges.  I've attached an audio clip of listening to a CW station.  The audio distortion is most prominent on dots, and less so on dashes.  I've also attached some spectrum plots of (a) the clip as a whole, (b) dashes, and (c) dots.  In the dot spectrum, I can clearly see that there are some additional artifacts that I assume are contributing to the sound I get.

Does anyone have any recommendations on how I can modify one or more of my filter sections to remove these artifacts?  I designed the filter using components I had on hand.  I don't have a schematic drawn up at the moment, but I'm looking for awareness of general principles I may have overlooked. 

I designed it using this page:  https://www.electronicshub.org/active-band-pass-filter/

I used the Multi Feedback Active Band Pass Filter for reference.  My values are:
C1 & C2 = 11.4 nF (specific values .01 uF metal film caps I had on hand)
R1 = 82k
R2 = 180k
R3 = 3.3k

Any thoughts greatly appreciated!

Regards,
-Rob KC4UPR


Rob French (KC4UPR)
 

Ah-ha.  Two new things... one, this is ringing.  I checked the spectrum on the dots a little more carefully, and I can see that those additional spikes are clearly multiples of the peak frequency.  So when a signal gets centered up in the filter, it has a tendency to generate the ringing. 

Two... this was heavily exacerbated by my AGC.  In retrospect, as one would expect (I think), this is is worse with AGC Fast, and better with AGC Slow... with fast, the abrupt change in amplitude is more likely to generate the rapid transitions that causes the harmonics.

That said, even with my AGC turned off, I can still find the harmonics/ringing, they just don't have the pronounced "thump" effect.  With AGC Off, they cause more of a buzz/resonance.

Still looking to figure out how to remove them.  In retrospect, my three section filter is probably not ideal... I'm considering trying to modify one of the sections to be a low pass filter to try to tamp down the harmonics.

Regards,
Rob KC4UPR


Rafael Pinto [PU1OWL]
 

Rob, 

Very sharp bandpass filters tend to ring. Also as you made 3 identical sections the phase margin gets really low, allowing the filter to act as an oscillator.

The thump sound might also be op amp instability because of the really sharp response.

Your filter is centered at 700hz, but what is your expected bandwidth? Maybe you can try with one section only and check whether the ringing and thumping improves. If that is the case then I suggest redesigning the filter as a 6th order Butterworth biquad cascade.


Curt
 

Rob

I appreciate the work here.  Pulsed signals like CW add another dimension, and Rafael offers good advice.  I won't necessary invite you to pursue the math involved (sophomore year EE that not all of us remember how to do). 

I also find even with a DSP filter - I tend to pick a particular favorite - 300 Hz for all my operating.  About the same for my K2 with its tunable analog filter.  A couple recommendations - review the hyper-mite filter (4 state qrp, NM0S).  Also if you really wish to explore - see if kc9on is still offering the CALF - its a nice board for experimenting with audio filters.  While my uBITX includes an adjustable SCAF, the fixed filters in the CALF easily sound better on CW (plus I 'designed' a couple for SSB).  I remember using a TI tool to design the filters - KC9ON likely has an article posted that describes how I did it.  (I would need to read it myself to repeat the work). 

73 Curt wb8yyy


Arv Evans
 


On Mon, Dec 28, 2020 at 9:09 PM Rob French (KC4UPR) <kc4upr@...> wrote:

Hello,

I've implemented a simple audio band pass filter for CW on my uBITX v5.  It uses 3 sections of a TL974 op amp running off of 5V regulated (the 4th section of the op amp is used a buffer for a line out).

The three sections are identical, and implement a bandpass filter with a center frequency of about 700 Hz (I've attached a screenshot of the audio response as I tuned across a CW signal on 40 meters).

The problem I have is that when a signal is tuned up in the passband, plus or minus a couple hundred Hz, I get a strong thumping or distortion effect on the leading edges.  I've attached an audio clip of listening to a CW station.  The audio distortion is most prominent on dots, and less so on dashes.  I've also attached some spectrum plots of (a) the clip as a whole, (b) dashes, and (c) dots.  In the dot spectrum, I can clearly see that there are some additional artifacts that I assume are contributing to the sound I get.

Does anyone have any recommendations on how I can modify one or more of my filter sections to remove these artifacts?  I designed the filter using components I had on hand.  I don't have a schematic drawn up at the moment, but I'm looking for awareness of general principles I may have overlooked. 

I designed it using this page:  https://www.electronicshub.org/active-band-pass-filter/

I used the Multi Feedback Active Band Pass Filter for reference.  My values are:
C1 & C2 = 11.4 nF (specific values .01 uF metal film caps I had on hand)
R1 = 82k
R2 = 180k
R3 = 3.3k

Any thoughts greatly appreciated!

Regards,
-Rob KC4UPR