Circuit modelling ? How do you know ?
- KVRAF
- 10426 posts since 3 Feb, 2003 from Finland, Espoo
Even with IC's you can at least get an idea by measuring input vs output, right? Circuit modeling comes in all forms though.. you could say you modeled a capacitor and just write the most basic filter.
"Wisdom is wisdom, regardless of the idiot who said it." -an idiot
-
Funkybot's Evil Twin Funkybot's Evil Twin https://www.kvraudio.com/forum/memberlist.php?mode=viewprofile&u=116627
- KVRAF
- 11559 posts since 16 Aug, 2006
What does modeling even mean? A model could be, "I approximated inputs and outputs based on measurements and made a model using traditional DSP." If that device was made with circuits, didn't I just model a circuit? Doesn't automatically mean it's a good or even accurate model. Could just be mimicking one setting and not even well. Doesn't not make it a model.
Don't get hung up on semantics or marketing. Demo gear. Check it out for yourself. Run tests. Use your ears. Buy what you like.
Don't get hung up on semantics or marketing. Demo gear. Check it out for yourself. Run tests. Use your ears. Buy what you like.
- u-he
- 28096 posts since 8 Aug, 2002 from Berlin
Regarding integrated circuits, for Repro we used the circuit diagrams shown in the patents filed for the respective VCO, VCF and envelope chips. Patents are long gone, the tech is available for anyone to use. Heck, some patents contains formulas which can be used 1:1 in code.
-
- KVRian
- 693 posts since 17 Sep, 2007 from Planet Thanet
Indeed, some of the people on diystompboxes and freestompboxes are experts. That's why things like the Klon were reversed
- KVRAF
- 23112 posts since 7 Jan, 2009 from Croatia
Well, there's no saturation in 32C's EQ "circuit emulation". So I find it highly dubious that they did any actual circuit modeling.mevla wrote: ↑Thu Nov 29, 2018 12:48 am It's a bit more certain when the software maker actually has constructed hardware devices. For instance the 32C EQ circuit emulation in Harrison's Mixbus32C DAW is claimed to be an exact replica of every resistor, transistor and other component of the 32C EQ from the famous Harrison mixing console of the 70s (Michael Jackson, Supertramp, Zappa, ABBA, Paul Simon, etc...) and that claim has more weight since Harrison made the hardware in the first place.
https://www.admiralbumblebee.com/music/ ... er.html#eq
-
- KVRAF
- 2418 posts since 9 Nov, 2016
Is there like a path one can go as a noob in this area to get a feel for it. I'd like to have a crack at trying to emulate a filter but I'm not a coder.Urs wrote: ↑Thu Nov 29, 2018 9:09 am Regarding integrated circuits, for Repro we used the circuit diagrams shown in the patents filed for the respective VCO, VCF and envelope chips. Patents are long gone, the tech is available for anyone to use. Heck, some patents contains formulas which can be used 1:1 in code.
So if gets too nerdy in terms of math and code, it's not for me. But if there is an intermediate level with some abstraction going on and some graphically oriented tools that can help you with generating formulas, it might be a nice project.
I often wonder how modeling is done.
-
- KVRian
- 693 posts since 17 Sep, 2007 from Planet Thanet
^
Have a look at this Reaktor video - it's a bit less math / code intensive than some (but it leads onto Vadim's book)
https://www.youtube.com/watch?time_cont ... pAxQIUGT4w
Have a look at this Reaktor video - it's a bit less math / code intensive than some (but it leads onto Vadim's book)
https://www.youtube.com/watch?time_cont ... pAxQIUGT4w
- u-he
- 28096 posts since 8 Aug, 2002 from Berlin
The first issue is the understanding of the relation between current and voltage. Once you grasp that, the maths are excessively simple.Stefken wrote: ↑Thu Nov 29, 2018 10:32 amIs there like a path one can go as a noob in this area to get a feel for it. I'd like to have a crack at trying to emulate a filter but I'm not a coder.Urs wrote: ↑Thu Nov 29, 2018 9:09 am Regarding integrated circuits, for Repro we used the circuit diagrams shown in the patents filed for the respective VCO, VCF and envelope chips. Patents are long gone, the tech is available for anyone to use. Heck, some patents contains formulas which can be used 1:1 in code.
So if gets too nerdy in terms of math and code, it's not for me. But if there is an intermediate level with some abstraction going on and some graphically oriented tools that can help you with generating formulas, it might be a nice project.
I often wonder how modeling is done.
The second issue is the implementation of these simple formulas in an algorithm. Simple formulas, extremely difficult problem.
-
- KVRAF
- 35569 posts since 11 Apr, 2010 from Germany
Yes.wagtunes wrote: ↑Thu Nov 29, 2018 12:14 am Does it really matter? I mean if the plugin sounds better than anything else you've ever heard that it's "supposed" to be modeling, who cares? If it sounds like crap, I won't buy it anymore. If it sounds great, I will buy it.
Isn't that all that really matters?
-
- KVRAF
- 2418 posts since 9 Nov, 2016
@resynthesisresynthesis wrote: ↑Thu Nov 29, 2018 11:06 am ^
Have a look at this Reaktor video - it's a bit less math / code intensive than some (but it leads onto Vadim's book)
https://www.youtube.com/watch?time_cont ... pAxQIUGT4w
Thanks for your answer. I realise that I asked for a graphical tool and Reaktor is an obvious answer.
Let me clarify more what I mean which a graphical tool.
I had a glace at mathlab and the story of the zeros, poles, and gain.
You can have graph representations and the numbers for a standard filter like e.g. a butterworth filter.
But I don't want to emulate a butterworth filter. How do you go about if you want to emulate a MS-20 filter for example. Are there tools where you can draw in graphs, or that can analyse audio and that will generate the numbers for you?
If this is a case of "you can't have your cake and eat it too" (it probably is) you can tell me so.
"If you're not willing to plunge in the deep and do the hard work, you can't get results." Fair enough.
But the math behind it, is pretty deep I guess.
Maybe what I need is a good documentary or a good book that shows the magic in an accessible way.
A documentary showing people measuring up equipment, digitally constructing the circuit diagrams and generating the functions and numbers for the dsp.
I thinks it is a thing of beauty when you see how people can digitally model an analogue device. It's just a pretty steep learning curve to understand it if you don't know where to start.
Make any sense?
Last edited by Stefken on Thu Nov 29, 2018 12:08 pm, edited 1 time in total.
-
- KVRAF
- 2418 posts since 9 Nov, 2016
Thans Urs. Are there accessible starting points you can direct me to. I don't know where to startUrs wrote: ↑Thu Nov 29, 2018 11:52 amThe first issue is the understanding of the relation between current and voltage. Once you grasp that, the maths are excessively simple.Stefken wrote: ↑Thu Nov 29, 2018 10:32 amIs there like a path one can go as a noob in this area to get a feel for it. I'd like to have a crack at trying to emulate a filter but I'm not a coder.Urs wrote: ↑Thu Nov 29, 2018 9:09 am Regarding integrated circuits, for Repro we used the circuit diagrams shown in the patents filed for the respective VCO, VCF and envelope chips. Patents are long gone, the tech is available for anyone to use. Heck, some patents contains formulas which can be used 1:1 in code.
So if gets too nerdy in terms of math and code, it's not for me. But if there is an intermediate level with some abstraction going on and some graphically oriented tools that can help you with generating formulas, it might be a nice project.
I often wonder how modeling is done.
The second issue is the implementation of these simple formulas in an algorithm. Simple formulas, extremely difficult problem.
-
- KVRAF
- 35569 posts since 11 Apr, 2010 from Germany
I'd think starting with a virtual Modular like Reaktor, SynthEdit, or Flowstone would be a good starting point. And with those, and a good documentation, you can even go down to programming level.
-
- KVRAF
- 6435 posts since 22 Jan, 2005 from Sweden
amp modeled circuit with 1 dB gain
output signal dB= input signal dB + 1 dB
out=in+1
that's all it could mean.
Ask if descrete components are modelled and see how much mumbo jumbo you get as answer.
output signal dB= input signal dB + 1 dB
out=in+1
that's all it could mean.
Ask if descrete components are modelled and see how much mumbo jumbo you get as answer.
-
- KVRAF
- 2418 posts since 9 Nov, 2016
I've glanced at Synthedit/Flowstone but I don't want to invest too much time in it as I don't see a future for it.
I've seen the dsp in Synthedit/Flowstone for your average, not so bad but also not so good, low pass filter.
There are also examples floating around on the web, but that's just copying some else's code.
But how do you go about to making a REALLY GOOD ms-20 filter; how do you make that repro filter.