Yup, looks like I misunderstood what this op-amp was doing, and it’s well inside the SVF structure. That said, it still might be worth putting a small (10 to 100pF range) ceramic cap in parallel with R19, between pin 1 and 2 of U3-A and maybe limit the bandwidth of it a little - see if that has any effect at all.
Cheers
No problem😁 sounds like a good idea to put this 100pf cap. Gonna test this as soon as i get home. Thanks
This is what a 100pf (101) ceramic cap over pin 1 and 2 of U3-A does
Even more resonance (with reso pot closed)
Maybe increase C9 22pf (U3-B) to 100pf will remove the spike?
Excuse my ignorance here, but doesn’t R32 (49.9K) provide resonance permanently in the schematic, I really don’t understand why it’s there if you have a resonance loop already. I would disconnect R32 and test your circuit.
The only thing I can think of is that its not a standard LP filter, but a state variable which looking at the schematics for most state variable filters there is some sort of resistor between input and output like this, so maybe ignore me.
Hi Craigyb, thanks for helping. Appreciate it. I assume you’ve read my post about this. Trying to eliminate resonance if reso pot is closed. I somehow have a little reso spike left as if there’s some bleedthrough or something. It normally wouldnt bother me but this one is to noticeable. Im gonna try bypass R32 and see if this changes anything. I’ll keep you updated. The 4 pole 24db version of the filter doesnt have this. I think it would be nice that the 2 pole 12db filter would do the same.
Don’t bypass it as this would be a wire, As I said the state variable is slightly different in design, but what I have noticed is these input and loop resistors are fairly consistent so 10k input, 10k to the output. Are you using the 1M input resistors or have you dropped the values to get higher sensitivity? Is it possible to increase R32 to match the input resistors or maybe use a 500K trimmer to see whats a good fit if it helps.
Yep, bypassed it and my scope throws lines off the chart, also the sound gets killed.
I pretty much used all values on the schematic. Only 49.9 doesnt exist i used 47k’s there.
Input resistors 1meg indeed. 100k is waayy too loud. 1meg fits the rest of my modular setup.
I will increase R32 to match value to input resistor. Will let you know asap. Thanks in advance for your help
I’m not surprised, double it 100K and see what happens
Yes i will try. Let you know asap
Uhm, i get some hi pass filtering on the square wave. Left top edge get rounded but only a bit. The resonance is behaving strange when R32 is 1meg. I tried also with 470k. Both not exactly right but it did round off the spike. My scope started to trigger so not sure if this is it. Sorry😁
Got to go now but i will test R32 to 100k tomorrow. Have a nice day. Bye bye
Bet it sounds good 
but I guess it’s the complete opposite of what we want - back to the drawing board I suppose…
Ok. I just did the following:
Changed R32 from 47k to 100k. This makes the reso spike smaller due some fair amount of volume loss. But the resonance is acting weird when turning up the resonance. It maxes out or something. Together with the cutoff pot it changes the waveform in a clean squarewave . Doesnt look normal and sounds off too.
To compensate for that volume loss i assume i could decrease the 1meg input resistors to lets say 470k or less, but with the risk that the reso spike bumps up too, leaving me with the same results. Right? Could i bump up volume without bumping up the reso spike? Or do i need to increase or decrease R19 (330k) resistor from U3-A to achieve this?
From what I’ve seen the input and that resistor are usually the same value, so if you drop the input resistance to 470k then then R32 should be the same
Hi, so if i change the 3 input resistors (R25, 31, 34) to 470k and also R32 to 470k, this removes the resonance spike? My goal is trying to get a clean square wave when no resonance is dialed in but if this is not possible because of the filter’s design (no zero resonance), then so be it.
Since the 4 pole 24db filter’s resonance doesnt have this spike, maybe we can find a solution and improve this filter as well.
The resonance circuitry in both of these filters (12dB and 24dB variants) is essentially a VCA made from U6-A. In the SVF it feeds an increased portion of the BP output back into the filter network. The control circuitry of the VCA is identical in both filters, so (for the moment), let’s assume that the issue is not that the VCA is not closing fully (if it were, we would expect the same spike in the 24dB LPF).
If the VCA is closing fully, then the spike is not caused by the resonance circuitry, but either by the filter network itself, or by where the signal is reintroduced into the filter network. To work out which of these we’re dealing with, let’s see what happens when you remove R22. This removes the resonance circuitry entirely. If the spike is still there, then we know that it is a product of the filter network. If it is gone then we know that it is caused by where the feedback is reinserted.
I have a hunch that the spike will disappear when you remove R22. I think that the feedback should be going to the input of U3-A, rather than U2-B. If the spike disappears when you remove R22, try connecting pin 2 of R22 to pin 2 of U3, instead of to pin 13 of U2.
Just read one of your earlier posts, where you mentioned that removing U6 did not remove the spike, so no need to remove R22.
In this case, we know that the spike is caused by the filter network itself. Try changing R26 to 39k, to match R28 (and the corresponding resistor in the LPF).
But notice that the gain of U3-C is quite high (about 5x). If we assume that the output level is ±5V, that indicates that the output of U4-A is about ±1V. In that case, for the LP level to match that of the input signal that it is being subtracted from, ~50k looks about right for R32.
Hi and good morning (here it is 07:19) and thanks for helping by looking into this.
Yes I’ve already have taken some steps (with help of other members here) to find the issue what’s causing this. If we could find a solution to either reduce this spike further, or that the resonance can be zero would be awesome. Its just little to much resonance. When I get home later today I will check R26 and let you know. thanks.
Uhm in an earlier post you said the following:
Just read one of your earlier posts, where you mentioned that removing U6 did not remove the spike, so no need to remove R22.
Ok.. so at this point i would love to minimize trial and error here as my pcb is as tricky as hell, so can we put this schematic in falstad or parts of it to test which component is causing this? Or is this an unrealistic thing to do? Never worked with falstad but im willing to do the extra mile here.
Hi all, i just removed R22 (150k), making sure that RB is connected to the 100k and 47k tied to -12, but unfortunately it didnt solve my issue.
No sound, no visual, no resonance (reso pot not working), only a little bit of high pitched oscillation.
So R22 is coming back right in as it seems a vital part of letting the audio pass.
