MUFF WIGGLER Forum Index
 FAQ & Terms of UseFAQ & Terms Of Use   Wiggler RadioMW Radio   Muff Wiggler TwitterTwitter   Support the site @ PatreonPatreon 
 SearchSearch   RegisterSign up   Log inLog in 
WIGGLING 'LITE' IN GUEST MODE

[IN STOCK] Eurorack: Discrete State-Variable VCF v1.4
MUFF WIGGLER Forum Index -> Music Tech DIY Goto page Previous  1, 2, 3 ... 49, 50, 51 [all]
Author [IN STOCK] Eurorack: Discrete State-Variable VCF v1.4
JanneI
I just finished my trio in custom 5U format. Works fine and sound great! I needed to build the notch buffers on veroboard, so here's a pic if someone is about to do the same thing. I used the 3x2 pin header ("LP-HP-GRD" with 2 pins removed) to mount the little veroboard to the main pcb.


elmegil
I"m trying to troubleshoot one of these.

Does anyone have:

a schematic for 1.4?

a schematic with designators instead of values (aside from the transistors)?

Thanks....
negativspace
I don't think I've posted one with designators... made up a quick PDF for you, though.
elmegil
Awesome, thank you very much!
2thick4uni
Reviving an old thread - I've had the pcb and panel for this great filter for a while now and am just getting back into SDIY so have finally starting to build it.

I had a glance at the schematic and the 220p capacitors in the feedback loops of the LP and BP output amplifiers looked way too large and can cause a considerable LP filtering effect in their own right. I've modelled this in LTSpice and in the case of the LP output amp, assuming 10v p/p output levels the frequency response is -12dB at 10kHz, and -18dB at 20kHz; so even with the filter fully open there is quite a bit of LP filtering going on here at the output stage which will dull the sound. This could be a design choice though, perhaps it sounds better that way?

This capacitor is usually installed to prevent instability which manifests as unwanted resonance at higher gains (the gain on the LP amplifier is x27), and a large value is needed here because the TL072 is not good in high gain applictions, it quickly becomes unstable and resonates. The solution is to use an opamp that is more stable at higher gain factors, so that a much smaller LP feedback loop capacitor can be used. Most of my experience is with SMD opamps, and the OPA1678 is very useful for this, I'll try some DIP op-amps and find a suitable part that will work with a much smaller feedback cap. Just swapping out the op-amp will have no effect unless the capacitor value is reduced too. If a 22p cap is used insted of 220p then the frequency responce will only be -1dB at 10kHz and -2dB at 20kHz.

I've attached a graph showing the cutoff effect of the LP amplifier (purple trace) and the BP amplifier (yellow trace). The frequency responce of the bandpass amplifier is slightly better due to it having a lower gain factor of x9, but would still benefit from better opamp and LP capacitor an order of magnitude or two smaller.

When I get it built I'll try different opamps and capacitors and report back on my findings.
zer0point
2thick4uni wrote:
Reviving an old thread - I've had the pcb and panel for this great filter for a while now and am just getting back into SDIY so have finally starting to build it.

I had a glance at the schematic and the 220p capacitors in the feedback loops of the LP and BP output amplifiers looked way too large and can cause a considerable LP filtering effect in their own right. I've modelled this in LTSpice and in the case of the LP output amp, assuming 10v p/p output levels the frequency response is -12dB at 10kHz, and -18dB at 20kHz; so even with the filter fully open there is quite a bit of LP filtering going on here at the output stage which will dull the sound. This could be a design choice though, perhaps it sounds better that way?

This capacitor is usually installed to prevent instability which manifests as unwanted resonance at higher gains (the gain on the LP amplifier is x27), and a large value is needed here because the TL072 is not good in high gain applictions, it quickly becomes unstable and resonates. The solution is to use an opamp that is more stable at higher gain factors, so that a much smaller LP feedback loop capacitor can be used. Most of my experience is with SMD opamps, and the OPA1678 is very useful for this, I'll try some DIP op-amps and find a suitable part that will work with a much smaller feedback cap. Just swapping out the op-amp will have no effect unless the capacitor value is reduced too. If a 22p cap is used insted of 220p then the frequency responce will only be -1dB at 10kHz and -2dB at 20kHz.

I've attached a graph showing the cutoff effect of the LP amplifier (purple trace) and the BP amplifier (yellow trace). The frequency responce of the bandpass amplifier is slightly better due to it having a lower gain factor of x9, but would still benefit from better opamp and LP capacitor an order of magnitude or two smaller.

When I get it built I'll try different opamps and capacitors and report back on my findings.


I'd be interested to hear how it sounds with these mods...I wonder if the extra high-end roll off is what gives this filter its "vintagey" character.

Maybe related - one issue with my unit is that at high input gain and resonance over 50% certain frequencies resonate in a harsh way that sounds like the microphonic feedback you get with certain guitar pickups - could that be the opamps?
2thick4uni
Quote:
I'd be interested to hear how it sounds with these mods...I wonder if the extra high-end roll off is what gives this filter its "vintagey" character.

Maybe related - one issue with my unit is that at high input gain and resonance over 50% certain frequencies resonate in a harsh way that sounds like the microphonic feedback you get with certain guitar pickups - could that be the opamps?


Not sure it will have much effect on the overall character of the filter itself as its only a first order (6db/oct) roll off, I'd have thought it would only really be noticeable as the filter gets towards fully open.

I don't think the microphonic feedback you've described is down to the output amps, the TL072 should be rock solid with 220p across it, however you can easily test it by soldering another 220p in parallel (which will increase the capacitance to 440p) and then check if this behaviour diminishes. This isn't a fix as it further reduces the bandwidth, just a test to see if it is the problem.
elmegil
Normally stability caps in feedback loops are 10p, 20p range....
cygmu
2thick4uni wrote:

This capacitor is usually installed to prevent instability which manifests as unwanted resonance at higher gains (the gain on the LP amplifier is x27), and a large value is needed here because the TL072 is not good in high gain applictions, it quickly becomes unstable and resonates.


Is that right? My beginner-level understanding of stability is that higher closed loop gain configurations are more stable rather than less -- hence the datasheets proudly claiming that an op amp is "unity gain stable", because low gains are harder to stabilise.

Have I misunderstood something? Or is there something special about the TL072?
elmegil
cygmu wrote:
2thick4uni wrote:

This capacitor is usually installed to prevent instability which manifests as unwanted resonance at higher gains (the gain on the LP amplifier is x27), and a large value is needed here because the TL072 is not good in high gain applictions, it quickly becomes unstable and resonates.


Is that right? My beginner-level understanding of stability is that higher closed loop gain configurations are more stable rather than less -- hence the datasheets proudly claiming that an op amp is "unity gain stable", because low gains are harder to stabilise.

Have I misunderstood something? Or is there something special about the TL072?


I would think that they'd be boasting about harder things being stable rather than easier things?

But rather than speculate, here is what Analog Dialogue has to say about stability:

https://www.analog.com/en/analog-dialogue/articles/techniques-to-avoid -instability-capacitive-loading.html

Looks to me like it's a combination of capacitive load and unity gain itself rather than the total gain? But I just skimmed the article.
cygmu
elmegil wrote:

I would think that they'd be boasting about harder things being stable rather than easier things?

That was my point -- hence "unity gain stable" is a boast and "gain of 1000 stable" is not.

Quote:

Looks to me like it's a combination of capacitive load and unity gain itself rather than the total gain? But I just skimmed the article.


Right. The situation you want to avoid is where the negative feedback becomes positive feedback. This happens when there is 180 degrees of phase shift at a certain frequency.
Frequency-dependent phase shift comes from the op amp itself and from capacitance and inductance so e.g. a capacitive load will introduce some, as will capacitance at the op amp inputs, etc.

It causes a problem if the loop gain at that frequency, i.e. the amount of signal that is fed back, is at least 1. But as the frequency goes up, the op amp's open loop gain goes down. The configuration will be stable if the op amp can't produce enough gain to feed back the whole signal at the 180 degree frequency.

Higher closed-loop gains (like the x27 in this circuit) come from delivering less feedback (1/28 of the signal in this case). So they tend to be more stable, because the op amp runs out of open loop gain at a lower frequency, before 180 degrees of phase shift arises.

Or so I believe! I would be happy to be corrected.

In practice for this circuit, those 220pF caps do seem to be too large, and it is really odd that the LP output has the output resistor inside the loop while the other outputs have it outside the loop. It's not wrong, but strange that they differ, and in the loop resistors are bad for stability, because when connecting to a capacitive load they give you an RC network which introduces the dreaded phase shift. Outside the loop, they separate the op amp from the capacitive load and stabilise things.
2thick4uni
Putting a capacitor in the feedback loop of an op-amp forms an RC filter with a LP function. The cut off frequency reduces as the as the capacitance increases, and the cut off frequency also reduces as the op-amp gain increases - if you have a look at the simulation I posted earlier it shows the difference in bandwidth between x9 and 27x gain - both are with a 220p loop capacitor. This is why we usually see small 10 to 22p caps in parallel with the feedback resistor, as it keeps the LP filtering effect above 20kHz which is considered the top of the audio range. Larger caps and higher gains can bring this down into the audio range, which is what I am seeing in my Spice simulation and in practice.

If an opamp that is unity gain stable becomes even more stable as the gain increases then these capacitors would be not necessary at all, but in practice this rarely seems to be the case. I think it depends upon the internal compensation and phase margin of the particular op-amp used and the capacitive loading on the circuit. I've found that the TL07x require considerable compensation in higher gain circuits, but this is based purely on practical experience. Also, a circuit may be quite stable without anything patched to the output, but if a long patch cable (i.e. a capacitor) is connected to it then it will go into oscillation and require more compensation.

As I say I only know the practical side of it - it will be interesting to have some comments from those with a better grasp of the theory.
cygmu
2thick4uni wrote:
f an opamp that is unity gain stable becomes even more stable as the gain increases then these capacitors would be not necessary at all, but in practice this rarely seems to be the case. I think it depends upon the internal compensation and phase margin of the particular op-amp used and the capacitive loading on the circuit. I've found that the TL07x require considerable compensation in higher gain circuits, but this is based purely on practical experience. Also, a circuit may be quite stable without anything patched to the output, but if a long patch cable (i.e. a capacitor) is connected to it then it will go into oscillation and require more compensation.


Stability definitely depends on the op amp's phase response and the capacitive loading etc. Those are the two main sources of phase shift around the loop. A third one, which might be important here, is the capacitance at the input of the op amp, which forms an RC network with the feedback resistor. I wonder if the problems you've seen with higher gain circuits come not from the gain as such but from the high value resistors? There's about 8pF capacitance at the inverting input of a TL072. Together with a 270k resistor that gives you 45 degrees of phase shift at 73kHz. This will not help.

The more I read and think the more I believe that higher gain makes things more stable. E.g. there's this quote from p19 of TI's article "Stability Analysis of Voltage Feedback Op Amps" https://www.ti.com/lit/an/sloa020a/sloa020a.pdf

Quote:

The original loop-gain curve for a closed-loop gain of one is shown in Figure 17,and it is or comes very close to being unstable. If the closed-loop noninverting gain is changed to 9, then K changes from K/2 to K/10. The loop-gain intercept on the Bode plot (see Figure 17) moves down 14 dB, and the circuit is stabilized.


I did some Spice sims to experiment, using the idea in TI "SPICEing op amp stability"

https://e2e.ti.com/blogs_/archives/b/thesignal/archive/2012/09/10/spic eing-op-amp-stability

The idea is just to apply a transient to the configuration and see how much it overshoots and rings. Anything over 20% overshoot is asking for trouble.

Here's what happens with a TL072 set up for unity (inverting) gain, no cap in the feedback, 470R in the loop, and a capacitive load of 200pF. I'm applying a 1mV transient so we are looking for -1mV output.



30% overshoot and a couple of rings.
Now here it is with gain of 27:

It's a bit slow to get there but no overshoot or ringing. So I do think that the increased gain stabilises things. I don't know if my model of the TL072 models the input capacitance, though.

Back to the unity gain case, with a 47pF cap added to the feedback:

Obviously much better than the first example

And finally, with the 470R outside the feedback loop, unity gain and no feedback cap:

which again is pretty good although it never reaches the target value because of the voltage drop across the 470R.

Quite enough from me! Apologies if you don't find this interesting -- I am totally drawn in by this kind of thing.
2thick4uni
Thats very interesting - I'm still waiting for some parts to finish the filter build, but I'll check this out when it arrives - looks like the first thing to try is removing the 220p caps completely with TL072's on the output amplifiers...
aragorn23
Hey all. Reviving this old thread yet again with a silly question: given that some of these pots are a bit hard to find, do you think it would just be fine to use 9mm alpha style pots instead and bend the legs to match the board footprint?
aragorn23
Hey everyone, which ferrites would you recommend for this build?
woodster
Earlier in the thread Jason (Negativespace) recommends this part from mouser - 623-2743001112LF

https://www.mouser.co.uk/ProductDetail/Fair-Rite/2743001112?qs=%2Fha2p yFaduhAVm6lzZG4AwTJqeaw8N6jPEik2NB%2Fqv%2FUb%252B2PdI9ESQ%3D%3D
aragorn23
Hi all. Just finishing up a build and I've noticed that on the v1.3 control board, the left most SMD cap and resistor are swapped around from the v1.4 control board (i.e. the resistor [4] is on the far left of the baord, whereas in the v1.4 pics the cap [C] is on the far left). Should I place these two components based on the 1.3 board labelling or is the labelling erroneous?
aragorn23
Hi all. Finished my build and I have some issues (I've calibrated already and it hasn't helped).

1) Notch and HP just have the straight input sound coming out of them regardless of freq and resonance settings.

2) Nothing is coming out of BP and LP, also regardless of freq and res. In fact the freq and res pots and cv inputs aren't doing a thing.

Any idea where to check? I didn't match the transistors so I'm hoping that's not the source of my woes...
MUFF WIGGLER Forum Index -> Music Tech DIY Goto page Previous  1, 2, 3 ... 49, 50, 51 [all]
Page 51 of 51
Powered by phpBB © phpBB Group