Circuit modeled filter, how to?

[djsubject]

OK,

Lets all talk about what we think is the best way to do what the OP wants!!!

instead of insulting each other

Xhip is shit hot for what it is, (unfinished)

these spore/strobe audio examples are shit hot (also to the best of my know how unfinished)

now, why the f**k you girls are all arguing about it Ben(KvR) only knows

every one behave or i'll start calling you all pussy's & get the thread locked

Subz

[nollock]

"Btw.. who specificaly do you mean?"

i talk to a lot of musicians and coders via various means, IRC, MSN, in person and so on. nobody you would know. i was not referring to shy, i do not talk to him outside kvr.

So why bring in the opinion of these people "nobody knows" if not to bolster your own position, an "appeal to authority" as WR put it?

Do you not realize how that looks?

"I know loads of people in the industry and they all think I'm L33T!!"

"Oh but its nobody anyone round here would know..."

<smirk>

andy seems to think users do not care about cpu use, perhaps not.

(specifically, he said cpu use was not the 'top priority' for most users by disagreeing with a statement i made to that effect..)

Some of them dont, or else "poly ana" which you mentioned, wouldnt have any sales. It absolutly twats my CPU, 2 or 3 voice poly at best. And after having a play with Orca I dont imagine Andys synths will be anywhere near as hungry as PolyAna.

All the rest of your argument about the quality of his results is pretty much irelivant because it's subjective opinion. I mean you can say in your opinion you dont think it's up to much, but other people have posted in this thread that it sounds awsome.

And to be perfectly honest this thread has gone so far now that I doubt you would even admit that it sounded great.

That's not meant to be a personal attack, just that most people rarely concede any ground if they can avoid it in arguments which get this contentious.

[nollock]

every one behave or i'll start calling you all pussy's & get the thread locked

Subz

Fook off u 1os3r! I want a bitchfight!!!!!!!!!!!!!!!

[Rock Hardbuns]

That's not meant to be a personal attack, just that most people rarely concede any ground if they can avoid it in arguments which get this contentious.

True that. It's also pretty useless. We are after all not a social club, no one is keeping score.

The initial argument was about what is more helpful to a beginner (which is actually pretty nice in and of it self) and then on to what is the better approach to circuit emulation. And then on to who said what to whom.. and why, and really who the hell cares?

( If I may suggest, if the combatants still feel the urge to prove them selves, create a new topic with a quick tutorial. Step by step instructions on how to derive code from an arbitrary DSP circuit, plus resource links and stuff.That's something the community values and remembers (regardless of whether it's the "best" way). )

Besides.. the bickering is starting to look sort of homo erotic.

[andy_FX]

But the principle might be the same: You tap the filter at different points and mix the taps in certain levels to harvest different filter responses.

Yep. And placing notches at exact frequencies is pretty easy too, in fact you can make the notches move around anywhere you want and could make that another parameter of the filter.

The Xpander/Matrix12 service manual is most helpful because there's a multiplexer for different configurations of resistors and inverters from each filter pole. From these configurations one can easily derive the levels in which the poles must be mixed to achieve the 15 filter types available in these machines.

Just a quick not on the expander circuit, since they can't tap the outputs anywhere they want they have to bypass the first cap and turn the filter into 3 cascaded one poles with feedback. The feedback is automatically calibrated when the switch is made. You need more feedback on a cascade of 3, which also gives you more low frequency dip as you increase the filter to self oscillation.

P.S.: Andy, I finally had a long play with Strobe last night and I think it's awesome! It's got that really creamy sound.

Glad you like it Hopefully the mod system will lead to all kinds of interesting presets.

[andy_FX]

1) take a sh-101, use it.
2) come back and listen to the spore demos.

Has someone done an sh-101 model called spore? Can you please post the links to the audio demos of this sh-101 model you are talking about?

For the record, I've written a synth called Strobe, which isn't a model of an sh-101.

[andy_FX]

http://vellocet.com/dsp/analog/SelfOscHarmonics.html

So basicly these give you an idea of what kind of saturation is in the resonance feedback?

Not really, you need the circuit for that, but once you have modeled the non-linearities in the entire filter you can double check against the harmonics.

Those harmonic spikes trailed down after the main resonant spikes are a result of distortion in the filter?

Yes.

So the moogs actualy have less 'in filter distortion' than the others?

Well the moog doesn't have a diode clipper to limit the self osc, it relies just on the Tanh like waveshaper generated by the transistors, which have a more gradual curve. The Tanh function comes from using perfectly matched simplified transistor equations.

The SH-101 is likely to be more assemtric clipping than the Juno 106.. as the former has a balance of even and odd order harmonics, the later more odd order harmonics?

Yes, for these particular examples generated from these particular synths.

[mystran]

Just a quick not on the expander circuit, since they can't tap the outputs anywhere they want they have to bypass the first cap and turn the filter into 3 cascaded one poles with feedback. The feedback is automatically calibrated when the switch is made. You need more feedback on a cascade of 3, which also gives you more low frequency dip as you increase the filter to self oscillation.

Does that cause havoc with tuning btw, moving the 180 degrees phase point that the feedback is supposed to rely on? I can't seem to match the 4053 stuff to datasheets reliably, but it looks like they swap the first cap for a 100pF one in reality, that I guess could help with tuning? Also, does all this make an audible difference other than increasing distortion slightly?

P.S.: Andy, I finally had a long play with Strobe last night and I think it's awesome! It's got that really creamy sound.

Glad you like it Hopefully the mod system will lead to all kinds of interesting presets.

Now I'm getting curious... is there an open demo somewhere or is it just few and selected individuals?

[aciddose]

"So why bring in the opinion of these people "nobody knows" if not to bolster your own position, an "appeal to authority" as WR put it? Do you not realize how that looks?"

yes i realize how it may have looked, but that wasnt what i was trying to say. i havent said "strobe sucks", or "nobody should use strobe" or anything of that nature. i've only said: "i do not think for the quality, although it is a step in the right direction, the effort invested was really worth it."

when i said "my judgment is backed up by highly skilled and experienced individuals" i was only talking about my certainty as to how i feel as an individual about the effort/quality issue. i've asked some other coders and musicians what they think, and we have all come to generally the same conclusion: it sounds good.. but it isnt really close to a IR3109 and in terms of over-all quality and cpu use it isnt all that much better than the macro-modeled brute force oversampling stuff like in poly-ana. judging by the rest of the sound, not just the filter, i didnt hear anything that really impressed me and i was in fact quite disappointed. again, i asked people i know what they think and they came to the same conclusions. i was only offering this as a justification for my personal impression, not offering it as an appeal to authority behind some stated facts; there have been no facts presented here, only opinion.

the argument was: "your opinions are wrong. andy's filter sounds better, therefore andy's methods are better and therefore andy's opinions are right." my rebuttal was: "my opinions are right in some circumstance, andy's filter doesnt sound good enough to me to justify using andy's methods, and andy is free to have any opinion he likes just as i am, neither of us are right or wrong."

of course, wr just wanted to troll and start flamewars here as usual, so we can clearly see his motivation was not to discuss opinions but rather to shoot one or the other opinion down without presenting any of his own.

facts are, i'll admit, if you want to make money andy's methods probably are better than mine. after all i give away my software for free, obviously not a wise money-making method.

[andy_FX]

we have all come to generally the same conclusion: it sounds good.. but it isnt really close to a IR3109 and in terms of over-all quality and cpu use it isnt all that much better than the macro-modeled brute force oversampling stuff like in poly-ana.

I'm still waiting for an email from you so I can give you a copy of Strobe, please check your private messages on kvr to get my email address. I'm not sure what you are commenting on in terms of cpu use, since you don't have a copy Strobe as far as I know.

I have not modeled the IR3109 chip, I don't even have a schematic for it (although if you have one please sent it too me, I would love to check it out), Strobe isn't a model of an SH-101, and I agree a real IR3109 analog filter sounds better than the model I have done. I still haven't optimised the filter, so what is there is c++ code only, no sse. Optimisation will be done later on in the dev cycle. I'm using a few different linearly interpolated non-linear tables that are run time generated from analog modeling techniques, as well as some low order polynomials where needed where components are never reaching their full non-linear regions. I could have use tables generated from qucs or something, but it's easy to tweek coeffs and generate several different tables at run time for added variation.

the argument was: "your opinions are wrong. andy's filter sounds better, therefore andy's methods are better and therefore andy's opinions are right." my rebuttal was: "my opinions are right in some circumstance, andy's filter doesnt sound good enough to me to justify using andy's methods, and andy is free to have any opinion he likes just as i am, neither of us are right or wrong."

My point has always been to simplify from the full equations / circuit measurements as you see fit. The math isn't that hard, and if it is you can just export something from qucs. The simplifications I have made for the Strobe filter are pretty much in line with what you have suggested previously to model an OTA based cascaded four pole, so what you are saying, in effect, is that your methods aren't good enough.

For filter cores, where you want to run many of them the tradeoff for cpu usage does limit the level of detail currently, but for effects like compressors, or if you are doing a dedicated filter plugin you can go into more detail in modeling the signal path. One thing I am keen to do is to have per voice different parameters for components like diodes and transistors generated from the users serial number, that way everyone gets a slightly different sounding synth

Andrew

[hifiboom]

One thing I am keen to do is to have per voice different parameters for components like diodes and transistors generated from the users serial number, that way everyone gets a slightly different sounding synth

thats funny.

does your circuit also change its soundquality while its aging inside the vst folder?


btw I thought your strobe thing is the orca synth. These are different things right?

[Rock Hardbuns]

does your circuit also change its soundquality while its aging inside the vst folder?

Have it drift out of tune after a while! For the authentic analogue experience!

[hifiboom]

hehe

but I see real value in this feature for a polyphonic synth, where every voice gets slightly detuned components not only the osc pitch but also the single elements of every voice signal chain ...
for a less static sound.

[andy_FX]

btw I thought your strobe thing is the orca synth. These are different things right?

Strobe is completely different dsp to Orca, different osc, different filter, different envs etc.

[andy_FX]

where every voice gets slightly detuned components not only the osc pitch but also the single elements of every voice signal chain ...
for a less static sound.

There is a mod source which has a different per voice per parameter value. This is not the same as varying every component, but at least you can get some variation. So you can effect with variable depth all parameters per voice through the mod system, that when you do an arpeggiation each voice will have it's own sound and give the arp a cycling rhythm.

There are also per voice per parameter random numbers generated on key on so you can do random mutations of every voice easily. There are loads of other cool mod sources as well.