Side Tracking Filters

[BobF]

BobsSideTrackingFilter.fsm

(210.07 KiB) Downloaded 1071 times

Hello gang,

This is a side tracking filter. Basically its tracks the center frequency of a incoming signal. For instance if you input a square wave and you have the middle filter set for low pass, you will get a sine wave of equal amplitude out as the square waves frequency is changed.

For more information, see full write up inside schematic.

Please if you find any cool uses for this or create any new configurations of your own, PLEASE post. Thank you all!

Later then, BobF.....

[noisenerd]

Ooh, this could be fun! Thanks Bob, I'll post if I come up with anything cool to use it in.

[Spogg]

Hi Bob

Sorry it's taken me some time to get to try this. I've been busy with my Quilcom FYB!

This is a very crazy but fascinating idea (as per usual)

I found some issues which I've tried to address and the Spogged version is here for you to have a look at.

This info below is also in a comments box in the Spogged schematic:
................
Spogged 6.6.16:

-Replaced envelope followers with fast rms converters for reduced waveform superimposition.
-Removed 1 pole integrator since not needed with 50mS rms window.
-Separate knobs now wired for dry and filtered for scope.
-Scope added to show the control signal.
-Since the control range seems to be non-linear and has an offset, I've added filter range and offset trimmers. These interact with the cutoff knob on the middle filter. It's a balancing act and this could be improved I'm sure. The filter schematic has some maths in it to change the range behaviour (originally, not by me!).
...................

In your original, the performance at frequencies below about 200Hz was kinda lumpy and thus had artefacts. This was due to the use of those envelope followers having ripple on them relating to the frequency (time constant issue). So the fast rms converters I used, set to 50mS window, behaved themselves right down to 20 Hz. This meant there was no further need for the integrator and removing it caused no loss of performance. You could also try peak followers where you can adjust the time constants and this may also give good results.

The other significant issue I spent some time on is the control signal for the middle filter. It was really hard to get the filter to track properly over the whole useful audio range. My solution is really a band-aid. I added range and offset controls for the filter cutoff control. When you change either trimmer you then need to adjust the cutoff knob again so you have 3 variables to set that all interact. In the upload I've got this set up to be not too bad but there must be a better way. This issue is not helped by the maths in the filter itself which affects the operating range of the filter, so it might be worth experimenting with alternatives.

Many thanks for sharing this. It's been a fascinating 3 hours looking at this!

Cheers

Spogg

[BobF]

Hello gang,

If your interested in this, definitely download this Spogg'ed copy.
Thanks to Spogg it tracks better, covers a larger range, and use less components.

Many thanks Spogg, I did not even have the "fast rms converter" in my tool box. Did you create it or did you find it somewhere? It really makes a difference.

Yes, I agree with you, if we could find a way to remove the the trimmers that would be great. This is where we need some real pro's to ring in hear. This is really the kind of feedback I like! Spogg spent some real time on this and I am very greatfull for his hard work.

As can be seen this is far from a completed module, but with people like Spogg and hopefully others it could become a useful module. Maybe better filters, different configurations, and etc.. One idea I have for a use, is taking a rich harmonic waveform like square or saw and tracking a given harmonic across several octaves. Hopefully others will come up with some other cool ideas. If so, PLEASE post.

Later then, BobF.....

[Spogg]

... Many thanks Spogg, I did not even have the "fast rms converter" in my tool box. Did you create it or did you find it somewhere? It really makes a difference.

You're very welcome Bob I do enjoy looking over your stuff.

Like so many of my toolbox items, I have no idea where I found this fast rms module but whoever made it is far more clever than me, so many thanks to the anonymous author. I used it in my Quilcom Mistress project as an alternative to the peak follower, which is also really good because you can set attack and decay separately whereas with rms you can just set a sampling window. As you know rms is for "power" and peak is for amplitude so the results can be different depending on the signal. In some instances you may want a fast attack and a long decay and rms can't do this (without further processing the output).

Cheers

Spogg

[martinvicanek]

Hey Bob, you remember that adaptive filter thingy I posted a while back here. It is a weird control loop, a bit tricky to make it respond fast yet keep it stable, because the quantity of interest is frequency. So you have to adjust your loop transfer function time constant to the operating point. I believe this is related and also applies to your schematic?

[BobF]

Thanks martin,

I will definitely look more into this. Thank you!

Later then, BobF.....