KVR Audio

rv0 · Post by **rv0** » Mon Nov 23, 2009 6:54 pm

samples gonna happen any time soon.. busy writing webapps, fixing synths, troubleshooting midistuff, trying to have a life.. aargh

in any case, sorry i'll get on it asap

antto · Post by **antto** » Mon Nov 23, 2009 8:07 pm

jobro: thanks ;]
rv0: don't worry! you are such a nice guy ;]
btw, is a MIDI-modded TB-303 gonna work the same way when only the sequencer is used?
i will probably have some money (we're selling an apartment) and i am looking around at ebay, i saw a TB-303 with a MIDI-mod, which was kinda cheap, probably because it is "modified"
but if this mod can still be switched OFF, and the synth works like _normal_ in all aspects - then i don't care about the mod
i have no clue if i will buy something, this is big money, i really want to have one, but there is a risk, i will probably regret it, altho, if it's working i guess i will be able to sell it for (almost) the same price..
any advice here is very welcome

rv0 · Post by **rv0** » Tue Nov 24, 2009 4:50 pm

i have midi on one of my 303's, should work as normal (there's a switch in the back)

i've seen midi mods installed in 303's with broken cpu.

I dont see a reason why a midified 303 would be cheaper?

don't spend more than $1000.. but it is a good investment if you are sure that what you are buying is in perfect working order..
but.. reality is, a 2nd hand 303 will almost always require servicing (like switch replacement)

antto · Post by **antto** » Tue Nov 24, 2009 6:00 pm

i was just thinkering and i figured i never asked the question: what happens if there is a tied note right after a slide?!

Code: Select all

G    G    O
c(S) c(U) c

first there will be the low C, on step2 the pitch will instantly go higher, but the low-pass filter will be activated and the output pitch will "slide"
but then on step3 there is no slide, the current pitch probably haven't yet reached c(U) at that time.. will it instantly "jump" to "c" or "slide" down?

Mike Janney · Post by **Mike Janney** » Tue Nov 24, 2009 6:52 pm

antto wrote:i was just thinkering and i figured i never asked the question: what happens if there is a tied note right after a slide?!
G G O
c(S) c(U) c
first there will be the low C, on step2 the pitch will instantly go higher, but the low-pass filter will be activated and the output pitch will "slide"
but then on step3 there is no slide, the current pitch probably haven't yet reached c(U) at that time.. will it instantly "jump" to "c" or "slide" down?

I'm not too familiar with your notation, but the slide would probably reach the target frequency on the second step unless you have very high tempo. The slide is activated on the third step aswell so if it hasn't reached it by the second step it will on the third.

antto · Post by **antto** » Tue Nov 24, 2009 7:38 pm

rv0 wrote:i have midi on one of my 303's, should work as normal (there's a switch in the back)

i've seen midi mods installed in 303's with broken cpu.

I dont see a reason why a midified 303 would be cheaper?

don't spend more than $1000.. but it is a good investment if you are sure that what you are buying is in perfect working order..
but.. reality is, a 2nd hand 303 will almost always require servicing (like switch replacement)

damn, you're probably right, what about a second hand x0xb0x then?
because the x0xb0x was "invented" way after the TB-303, i suppose it has stronger knobs and switches?
as far as i can tell, the x0xb0x is the closest hardware thing to the 303, even if some parts are slightly different, at least the circuit is the same, so the behaviour must be similar enough to be considered a TB-303
if i ever buy a TB, or a x0x, it'll be for scientifical purposes ;P~
erm, i'm not saying i won't have FUN with it ;]

Mike: well, as far as i know (from rv0) there is Slide and Tie
Tie just extends the duration of a note (and it is programmed in Time Mode)
Slide does the same, but it also takes another pitch from the memory (and thus it's programmed in PitchMode)

i'm using rv0's notation where:
G < note ON
O < tie
- < rest

and the second line was pitch mode, where the Up/Down/Accend/Slide are written in the brackets next to each pitch value

so every time there is a "G" in time mode (new note) the sequencer takes another pitch value from the pitch memory
and thus, a "tie" doesn't take one
this also means that you cannot have an accent on the "tie"
since the Accent/Slide information is written with each pitch value
kinda confusing, rv0: correct me if i said something wrong here ;]

now whenever there is a slide on a step, the step is extended (just like a normal "tied" step would be)
but on the next step, the pitch is changed, and the "slide" is actually some sort of a 1-pole LP filter circuit which is "switched" at a low frequency (an equivalent of a 7.5Hz LPF)
most of the time (because the tempo is very fast) the final pitch is never reached

also, if there were several slided notes one after the other - this LP filter doesn't get switched back, it continues from where it was, simply the sequencer is changing the pitch but the filter is at 7.5Hz and things are smooth
now if at any time there is a step without a Slide, this LP filter is switched back to it's normal frequence (which is very fast, i consider it's instant)

my question is, what happens when there is a slided note with a tied note right after it?

it will look like two slided notes, but the second won't have a pitch change
but because in a "tied" note, there is no "slide" i guess the LP filter will switch to the normal frequency which will result in a long note with an "instant" pitch change, which i think i haven't heard yet?!
hm..

Mike Janney · Post by **Mike Janney** » Wed Nov 25, 2009 7:59 am

antto wrote: so every time there is a "G" in time mode (new note) the sequencer takes another pitch value from the pitch memory

and thus, a "tie" doesn't take one
this also means that you cannot have an accent on the "tie"
since the Accent/Slide information is written with each pitch value
kinda confusing, rv0: correct me if i said something wrong here ;]

Not exactly, what happens is if you try programming two steps, C and C and then slide on the first step what the cpu/memory will store is actually a C with a tie on it. This is definitely one of the quirks of the 303 sequencer, where it actually _removes_ one of your programmed pitch memories. I don't know how this is represented internally in the 303 as no one (as far as I know) has disassembled the software yet. The reason you can't set an accent on a tie is because you set the accents in pitch mode. If you programmed a C and C with accent, then slide on the first C it would actually do exactly that (and not convert the C-C to a tie!). Weird huh? Also, the accent signal stays high during subsequent slided and tied notes via a latch (IC 13). So this part is actually done in analog (well digital) and not via software.
In reality there are actually *two* types of 303 accents, which I'm sure you've seen Antto. There is the totally smooth discharging accent (which also keeps the VCA open after the note gate is shut), and then the shorter which you sometimes can see as a dicontinuous jump in the VCA level. The second accent has more bark to it and does not "echo" after the gate is closed.

now whenever there is a slide on a step, the step is extended (just like a normal "tied" step would be)
but on the next step, the pitch is changed, and the "slide" is actually some sort of a 1-pole LP filter circuit which is "switched" at a low frequency (an equivalent of a 7.5Hz LPF)
most of the time (because the tempo is very fast) the final pitch is never reached

my question is, what happens when there is a slided note with a tied note right after it?
it will look like two slided notes, but the second won't have a pitch change
but because in a "tied" note, there is no "slide" i guess the LP filter will switch to the normal frequency which will result in a long note with an "instant" pitch change, which i think i haven't heard yet?!
hm..

The analog section of the 303 can't distinguish between tie and slides really. Think about the analog section as decoupled from the computer controller. All the control inputs you really have to the analog section are Pitch, Accent, Gate, Slide. Pitch is a CV signal, and Accent, Gate and Slide are digital signals (either on or off). So if you take care and simulate all 8 combinations that can occur (maybe not programmable via the sequencer though) then your simulation will be equipped to realistically reproduce any sequencer you can imagine being programmed, regardless of whether is via midi, the 303 sequencer or other means.

--Mike

antto · Post by **antto** » Wed Nov 25, 2009 4:35 pm

hmmm..
so yes, a slide of C->C is automagicaly stored as a "tie" ... i knew that

i was wondering what happens if you have a slide and then a "tie" ?
i probably gotta draw a picture because i'm not sure if i express myself clearly (grrrr)
http://www.box.net/shared/imkxtfnpj3 <- image
ok, two cases
the first case is where there is a single slide (C>F) and a normal note after that (F)
tempo is very fast, i draw the pitch if the filter didn't "switch" .. it would continue rising till it reaches F, but because there is a normal note, the filter is switched and there you go, F

case2: i'm not even sure if this is correct
a normal slide which should continue with a tie
the result should be a long note (taking 4 steps length) tho, the last two steps are the "tie" which is (F)
hm.. i still don't know what happens ;]

antto · Post by **antto** » Wed Nov 25, 2009 5:10 pm

Mike Janney wrote:
antto wrote: so every time there is a "G" in time mode (new note) the sequencer takes another pitch value from the pitch memory

and thus, a "tie" doesn't take one
this also means that you cannot have an accent on the "tie"
since the Accent/Slide information is written with each pitch value
kinda confusing, rv0: correct me if i said something wrong here ;]
Not exactly, what happens is if you try programming two steps, C and C and then slide on the first step what the cpu/memory will store is actually a C with a tie on it. This is definitely one of the quirks of the 303 sequencer, where it actually _removes_ one of your programmed pitch memories. I don't know how this is represented internally in the 303 as no one (as far as I know) has disassembled the software yet. The reason you can't set an accent on a tie is because you set the accents in pitch mode. If you programmed a C and C with accent, then slide on the first C it would actually do exactly that (and not convert the C-C to a tie!). Weird huh? Also, the accent signal stays high during subsequent slided and tied notes via a latch (IC 13). So this part is actually done in analog (well digital) and not via software.
In reality there are actually *two* types of 303 accents, which I'm sure you've seen Antto. There is the totally smooth discharging accent (which also keeps the VCA open after the note gate is shut), and then the shorter which you sometimes can see as a dicontinuous jump in the VCA level. The second accent has more bark to it and does not "echo" after the gate is closed.

uh? i know of two types of accent
accent (a new note starts with an accent) and an accent after a normal-slide (a non-accented note is slided to an accented)
the second might be fully "hidden" when tempo is slow, and the Decay knob is low.. when tempo gets faster and decay knob is high, the accent start to look like normal accents (higher amplitude and "gimmick")
are these the two types you had in mind or have i missed something?

rv0 · Post by **rv0** » Wed Nov 25, 2009 6:44 pm

antto:
i think i have already given you an example in one of the recordings i did
but anyway, i'll record it once i get to recording stuff
just check it out, i've done a few stuff with slides like that i think

- about the 2 accents types?
i dont understand either.. but the accent has a volume knobs so that will perform what is described?

- about x0xb0x components being better quality:
total BS.. it all depends on who's making it, + the 303's components are very good quality. all ALPS switches and pots.. It's a miracle these still work.
The ALPS switches now have a dustproof replacement that is so good, it will survive you and me. All other similar switches are crap in my experience.
also, the x0x gets the sequencer all wrong so you wont be getting exactly what you wanted. i hated it myself.. also hated the feel of knobs and buttons but maybe i'm just biased.. there's also so hard-to-get-by parts you need to emulate closer
oh and the x0x uses another pcb layout so that will color the sound character of the original.. if you wanna emulate the original, use the original.
- to x0x-owners reading this: please don't start a discussion about it here..

antto · Post by **antto** » Wed Nov 25, 2009 7:51 pm

yeah, you're right about this, even small variations on the "pcb" will definately make differencies..

uhm, do you know someone who's selling a TB-303? ;]
i think i don't mind about scratches, as long as it's working..

rv0 · Post by **rv0** » Wed Nov 25, 2009 9:53 pm

the allmighty samplepack:

http://users.pandora.be/darffader/rvooh ... 251009.zip

get it while you can, sorry for mistakes, explanation txt included!

damn i just notice its in the wrong directory..
both of you let me know when you have the file so i can delete it

and 303's for sale: well, they show up regularly but not cheap, always about 950 euro..
i remember buying my first 303 for half that price, and even that was expensive.

Kriminal · Post by **Kriminal** » Wed Nov 25, 2009 9:58 pm

Robin from www.rs-met.com wrote:edit 2009/10/14: project name changed to 'Open303'

as announced in another thread, i'm going to start a free and open source project for a TB-303 emulation plugin. the plugin should also include the Devil Fish modifications and some useful extensions on its own. the basic synth architechture is in place and now it's about time to go about tweaking the various internal parameters of the underlying signal processing algorithms. and that's where the collaborative nature of the project comes in: i need your help, if you:

-have some knowledge about internal technical details of the TB-303 / Devil Fish
-have access to te real thing and are willing to provide audio files for analysis
-are a graphic artist and willing to contribute to the GUI (at the moment, the plugin is GUIless)
-a programmer who wants to actually implement a GUI (i'm only familiar on making GUIs with JUCE, but possibly JUCE is not the optimal framework here - dunno)

if you think, you can contribute in other ways, that's of course very welcome as well. i hope, that we can use this thread here as kind of development blog, to discuss the various implementation details, analyze the real thing, tweak the code, etc.

O.K. - nuff said - here's the very first prototype:

www.rs-met.com/temp/AciDevil.zip

the archive inlcudes all the source, a project/solution file for MSVC 2008 and a compiled .dll for instant fun.

Let's make proper acid available for all!

peace, Robin

any progress on that Robin?

rv0 · Post by **rv0** » Wed Nov 25, 2009 10:09 pm

Kriminal wrote:

any progress on that Robin?

he posted the code repository a few page ago.

What I was missing though was a wiki feature, or maybe I didn't check well..
Like the pattern notation thing has been explained here like 20 times already

antto · Post by **antto** » Wed Nov 25, 2009 10:22 pm

i got em
thank you!

the tempo.wav is real good, gonna check it out tomorrow in detail
at first look over the 88200 vs 44100Hz samples, it seems the HP filter in the AD convertor is at a fixed frequency (which is logical, and good in my case, i needed to be sure)

Robin seems to be silent, but whenever he shows up i'm able to send him the samples
i guess you can remove them..

Open303 - open source 303 emulation project - collaborators wanted