Getting a more 303 sounding box..

[rarara]

Hi, the discussions around the new TT-303 and how authentic it is/isn't have made me think that I should put some work into getting my x0xb0x sounding as close to a 303 as i possibly can. I have one of the early runs and believe from memory it has a BA6110, and has since had the Bcbox i/o mod added. I have calibrated the machine in terms of frequency/TM4/TM5 etc and that helped, but still feel the sound has a bit more to do.

Looking through the threads I can see that the consensus is that BA6xxx doesn't make a huge difference, but there is a lot of talk about certain components (Q8, Q9 etc etc) having the correct values.

Is there any kind of list of the most important components to check, and what their values should be??

I can see there is an eagle file which lists voltages, but I am not interested in fixing any faults - purely in improving the sound. Can anyone advise. Also, I can see talk of matching certain components - say using a bunch of high beta resistors etc - can i test my current components in-situ to see how they compare?

thanks

[antto]

as i already did some tests about Q8 - it makes no difference, not just the hFE value but even the the transistor itself (i found 9 different types of PNP transistors from my old VCR which all worked and sounded the same)

TM4 and TM5 won't make your x0x more 303-ish since these are for tuning your VCO, however, if you slightly detune the "width" (TM5) you can make certain patterns sound a bit more interesting
i think probably in quite a number of acid tracks, the 303s that have been used were "off" more or less

TM3 is a more interesting adjustment, it shifts the frequency range of the filter (so it's equivalent to adding offset to the CUTOFF knob)
many people immediately say that "these two 303s/x0xes don't sound the same AT ALL" if only this setting isn't equal on both units, so it makes a huge difference altho it's a very simple thing

C29, as was discovered not long ago, on all TB-303s this capacitor is soldered "backwards"
it doesn't make a huge difference tho, i suggest you socket it...

now.. for making the x0x sound more like a 303, you need a reference, do you have a TB-303 there?

i have been doing this thing since i got my x0x, and while i don't have a TB-303, i have audio recordings from darffader's 303, which i use as reference
so i did some adjustments according to that

1) my envelope minimum decay time was slightly longer (this is the decay time on accented notes or when DECAY pot is at minimum)
this decay has an impact on the shape of the "wow" on accented notes
i desoldered R138 and used a pot to be able to temporary change it's value while looking on spectrogram to compare the decay time vs the reference, when i nailed it - i measured the resistance of the pot and used 2 resistors in series that gave the same value

2) resonance adjustment - i really recommend this to everyone
in my case i simply wired up a 10K pot in R97 because i need to be able to bring the feedback up to crazy levels for scientifical purposes
but for normal usage i would say you probably only wanna adjust the R97 value between 10K and 8K (below 8K it becomes really insane)
so maybe use a 2K trimpot with 8K resistor or something

i also socketed some other components
C13 - this acts as a LPF and creates the "wow" on accents
Q8 - square waveshaper transistor, tho it makes no difference what i put in there
C11 - square waveshaper this DOES have an effect, it is responsible for the PW characteristics of the square
R36 - square waveshaper this DOES have an effect, it can offset the PW of the square
Q28 - sawtooth buffer, i have placed a J201 tho it didn't change anything
IC11 - ehm, this doesn't have much to do with the sound
Q9, Q10, Q11 - these do some voodoo and are responsible for the VCF cutoff frequency and the parameter "coupling" between the cutoff/envmod/accent, haven't tested them much but basically the act as amplifiers for the signals i think.. Note: i haven't socketed Q11 actually

[isak]

Hi Antto old good stuff you posted here
Don't forget to add your list this cool old post by 3phase
viewtopic.php?f=12&t=10868&hilit=Styroflex
OMG...reading my old posts make me laugh on my self.
good times we had back then, I miss thous times.

[antto]

isak: interesting, i only read the first page roughly
too much talking and swapping components.. not very useful
and most importantly - no audio recordings..

one thing i noticed, he said about Q8.. wtf.. according to my tests, you can put a number of alternative PNP transistors with all kinds of hFE and it still sounds the same

oh well.. whatever

[antto]

i read thru the whole thread now..

my impression is that 3phase talks about component like transistors with words like "smooth" "harsh" "rounded" and as this applies no matter in what place they are used (be it in a 303 or an 808 or a TV)
kinda similar to the way he talks about capacitors..
he also was testing different capacitors in C21 and C20 expecting to hear some magical difference
this clearly shows that he simply doesn't understand (nor is interested to know) what happens in that circuit
because then he would know that those two caps are not involved in the resonance, and cannot be blamed for anything

the only useful thing in that thread was bxbox's test with the power supply
..and his audio recordings - wtf there is something so wrong with his 303, normally the accent shouldn't do like that, that happens if his env decay is wrong (long), or if he has shorted out r46..
besides that, i didn't see/hear any difference
also, i think he either didn't take the audio from the output jacks (but from somewhere inside the synth) or he had some other gear in the signal path (mixer or something)

someone there made a good point tho
this myth with the "use all HIGHEST BETA transistors" probably means that almost all x0xes have high beta transistors
in some places the Beta value will not matter, and in the places where it does .. well maybe the 303 in practice has low-ish beta instead

i question the way that guy compares his x0x with his 303
if he expects to "hear" the same thing when he sets the knobs to point to the same positions - that's so naive
even pots themselves have a "response curve" (especially those "log" ones)
this curve doesn't change the sound, it only changes the parameter response

i gotta fire up my x0x these days and test Q9/Q10 (since i have a solid reference for realtime A/B tests)
i expect that the transistor responsible for the envelope modulation would make a significant difference (the beta value) it would basically scale the effect of the envelopes

[antto]

lemme just point out something very important here:
the question "how to make my x0x sound like a 303" cannot be answered untill you answer the question "how does a 303 sound like"
IMO you first need to get a 303 as a reference, then compare the differences, understand what these differences are and what causes them
blaming the component brands, shapes, colors, age, and swapping them blindly simply won't get you far

and please, if anyone is reading this and if he will attempt to do comparisons, don't use those silly words (warm, cold, phat, thin, dirty, clean, gritty, oily, pink, smooth, yellow, edgy.. etc. etc..) to describe stuff
just record proper audio material and link it, then describe the conditions under which you did the tests
my understanding of "smooth" could be different ya know ;]

[isak]

and please, if anyone is reading this and if he will attempt to do comparisons, don't use those silly words (warm, cold, phat, thin, dirty, clean, gritty, oily, pink, smooth, yellow, edgy.. etc. etc..) to describe stuff

you killed me!!!!
You remind me the Tarantino movie "From dusk till dawn"

‏http://www.youtube.com/watch?v=P3pmwrtlDaY&feature=youtube_gdata_player

the question "how to make my x0x sound like a 303" cannot be answered untill you answer the question "how does a 303 sound like"
IMO you first need to get a 303 as a reference, then compare the differences, understand what these differences are and what causes them
blaming the component brands, shapes, colors, age, and swapping them blindly simply won't get you far

Totally agree with you here, even if you take a303 and test it with another 303 they won't be the same sounding.
I dont believe Roland checked each transistor betas before soldering them.

[antto]

i just tested Q9 and Q10 (since they are socketed)
i have those same bunch of transistors that worked in Q8 without any difference (they are of different kinds, and different hFE values)

i cannot measure the hFE cuz they have short legs (since they are pulled out of my old VCR) ***
Q9 makes no difference

Q10 (as i expected) scales (acts as an amplifier) the whole CUTOFF+ENVMOD+ACCENT control signal
so it's kinda the same thing that TM3 does (and thus can be compensated)

*** EDIT: took some brand new transistors (with long legs) and measured them
in Q10:
A1015Y - 199 hFE - control signal is shifted up
115 W2E - 256 hFE - control signal is shifted down
so the Beta value of the Q10 transistor makes the combined cutoff frequency signal scale up as Beta is lower
this means that if all x0xes use "high beta" they have a better chance of not being able to reach high cutoff frequencies

[rarara]

hi antto, interesting stuff....

with TM5, if it only makes some patterns better do all the others sound the same or in any way worse?

TM3 - i think i previously turned this fully clockwise to get the resonance mod (x0xio) sounding better. is that sensible or will that make any difference to other parts of the overall sound?

envelope minimum decay R138 - is it possible to nail the sound by ear (spectrogram?), and would there be any worth in leaving R138 as a pot?

resonance adjustment R97 - think I may do this mod...

C13 socketed - anything specific in terms of type/beta works best, or just randomly swapping?

C11 - same question as above?

R36 - same..

Q9, Q10, Q11 socketed - again, just get a bunch of alternatives and swap at will? most threads have stated high beta of 8,9,10 and possibly low beta of 11 (seen mention of swap 526 to 945 - need to check what I have) are best. confusing

I still seem to be seeing a lot of conflicting info about other components though:

TM6 - been mentioned that going from 5.33 to 5.84v works well. anyone care to comment?

still leaves the following C19,C20,C21,C24,C25,C26,C27 - am i correct in thinking that foil caps for these are difficult to source

as i already did some tests about Q8 - it makes no difference, not just the hFE value but even the the transistor itself (i found 9 different types of PNP transistors from my old VCR which all worked and sounded the same)

TM4 and TM5 won't make your x0x more 303-ish since these are for tuning your VCO, however, if you slightly detune the "width" (TM5) you can make certain patterns sound a bit more interesting
i think probably in quite a number of acid tracks, the 303s that have been used were "off" more or less

TM3 is a more interesting adjustment, it shifts the frequency range of the filter (so it's equivalent to adding offset to the CUTOFF knob)
many people immediately say that "these two 303s/x0xes don't sound the same AT ALL" if only this setting isn't equal on both units, so it makes a huge difference altho it's a very simple thing

C29, as was discovered not long ago, on all TB-303s this capacitor is soldered "backwards"
it doesn't make a huge difference tho, i suggest you socket it...

now.. for making the x0x sound more like a 303, you need a reference, do you have a TB-303 there?

i have been doing this thing since i got my x0x, and while i don't have a TB-303, i have audio recordings from darffader's 303, which i use as reference
so i did some adjustments according to that

1) my envelope minimum decay time was slightly longer (this is the decay time on accented notes or when DECAY pot is at minimum)
this decay has an impact on the shape of the "wow" on accented notes
i desoldered R138 and used a pot to be able to temporary change it's value while looking on spectrogram to compare the decay time vs the reference, when i nailed it - i measured the resistance of the pot and used 2 resistors in series that gave the same value

2) resonance adjustment - i really recommend this to everyone
in my case i simply wired up a 10K pot in R97 because i need to be able to bring the feedback up to crazy levels for scientifical purposes
but for normal usage i would say you probably only wanna adjust the R97 value between 10K and 8K (below 8K it becomes really insane)
so maybe use a 2K trimpot with 8K resistor or something

i also socketed some other components
C13 - this acts as a LPF and creates the "wow" on accents
Q8 - square waveshaper transistor, tho it makes no difference what i put in there
C11 - square waveshaper this DOES have an effect, it is responsible for the PW characteristics of the square
R36 - square waveshaper this DOES have an effect, it can offset the PW of the square
Q28 - sawtooth buffer, i have placed a J201 tho it didn't change anything
IC11 - ehm, this doesn't have much to do with the sound
Q9, Q10, Q11 - these do some voodoo and are responsible for the VCF cutoff frequency and the parameter "coupling" between the cutoff/envmod/accent, haven't tested them much but basically the act as amplifiers for the signals i think.. Note: i haven't socketed Q11 actually

[antto]

well if you read what i said, you first have to figure what's "wrong" with the sound of your x0x
if you can't tell by yourself, then just do an A/B test and upload it, we can then all have a listen and figure out where the difference is, and how to "fix" it

TM6 - this adjusts the 5.333 voltage which is used all around the circuit, specifically in the filter and oscillator
if you tweak it and make the 5.333 voltage more - a few things will start to change
1) VCO Pitch will go off, you'll have to retune it again
2) CV (Pitch Out) will also go off, but you cannot "tune" this one, since it is meant to work with exactly 5.333V, if you don't use the CV/Gate outputs then you might not care about this
3) VCF cutoff frequency will shift
4) VCF resonance will be increased
5) VCO Square wave will be affected, it will start to change it's sound a bit

there could be more side effects but these are the most significant ones

there are alternative ways to control those things i listed, so you better just leave TM6 calibrated to 5.33V


TM5 - well people use guitar tuners, oscilloscopes, or other tools to perfectly tune their VCO (the octave scale)
that's all fine, but sometimes i kinda like how it sounds when it's a bit off..

[rarara]

well if you read what i said, you first have to figure what's "wrong" with the sound of your x0x
if you can't tell by yourself, then just do an A/B test and upload it, we can then all have a listen and figure out where the difference is, and how to "fix" it

its for this reason i'm looking for things/values that are quantifiable (beta, voltage, resistance etc) rather than subjective descriptions of sound. i'm interested in things i can measure within my x0xb0x that the consensus is have an effect on the sound

TM6 - this adjusts the 5.333 voltage which is used all around the circuit, specifically in the filter and oscillator
if you tweak it and make the 5.333 voltage more - a few things will start to change

does a 303 get 5.33v exactly, with the exception where there is some power supply variation?

[antto]

you cannot ignore the effect of the passive components, they also have tolerance
no two 303s will sound identical because of that
but they will behave the same way because it's the same circuit

what i'm saying is.. i cannot tell you what components to put in your x0x
what i can do is.. if you do an A/B test with a reference TB-303, i can (try to) tell you what the most significant difference is, and which components you need to play with in order to minimize the difference

this is what i've been doing myself, as already mentioned, i have a number of high-quality dry recordings from darffader's TB-303 which i needed for my softsynth project
i used them as a reference for my x0x

does a 303 get 5.33v exactly, with the exception where there is some power supply variation?

the trimpot is there in order to make it possible to calibrate the 5.333V voltage precisely
because both the x0x and the 303 rely on this exact voltage to make the CV (Pitch-Out) exactly 1.0V/Oct and not 1.01V or 0.99V per Octave

[rarara]

swapped R97 with a mini pot and liking the change already. still more to do...

been reading a fair bit about the 303 clock/sequencer being less stiff and mention of gate per note of 5.25. could this side of the TB behaviour be replicated? i know the mario OS allows specifying a gate time but is that in any way comparible?

[antto]

clock being less stiff is not true
the 303 internal clock is an analog square wave oscillator, no voodoo there, no magical shuffle, nothing

the thing about "5.25" gate length - i might have said such a thing long ago, that's b*llshit, sorry
the gate length is determined by the clock, which as i already said is a normal square wave osc
the only other thing is the VCA envelope release (it's the time it takes for the capacitor responsible for it, to discharge)
should be the same as on a 303

[rarara]

vca envelope release - what in particular on the 303 should we be trying to replicate?