TB-303 'Facts' a Myth?

[ENV1]

I really don't see how playing around with z3ta+ is going to reveal anything about the inner workings of the 303.

Well nobody said that it would. Like i said in the first post, i used z3ta+ merely to visually demonstrate the effect the HPF has on the waveform since its waveshaper happens to have a waveform display.

Just look at the schematic... The answers are all right there!

I intend to. But since others have no doubt already done that, why shouldnt i avail myself of the knowledge that others have already gathered. After all there are only 2 possibilities here; either im right or im wrong.

[antto]

Thanks. I tried to check this out but the site doesnt open in my browser.

I take it its some kind of circuit simulator?

yes, it's a java applet circuit simulator..

first of all, the saw is pretty much an ideal ramp wave..

You mean the raw oscillator, right? Because i know for a fact that the audible 303 Saw is nowhere near a perfect ramp, (far from it), it looks very much like in the clip in the first post after applying the 'Window' parameter. (Not 100% like that, but the difference is negligable.)

you are not hearing the audible saw.. you are heary an almost perfect saw, which goes thru about 10 highpass filters, a ladder filter, a VCA, and a mixer

there are like 9 more highpass filters afterwards, including 4 or 5 in the actual filter (and its feedback path) some of which are tuned much higher than they should be for DC-blockers, and that's why the output waveforms look bent like crazy

I think there can be no doubt that it is the highpassing that is ultimately responsible for the odd waveshapes. After all when you test this you can clearly see that a HPF is always giving exactly this unique result, thus the probability that this could be mere 'coincidence' is practically zero.

if you feed almost any other harmonically rich signal into that same signal chain, it will look weird too.. that's not something unexpected, but it won't sound like "303" .. it will just sound a bit EQ'ed maybe

the highpass doesn't contribute to the square wave, it contributes to the lack of resonance in when the filter cutoff is low, in the bassy frequencies..

Well it 'contributes' at least in so far as that it is the decisive factor with respect to the waveforms ultimate shape, no?

the few highpass filters in the actual feedback path of the ladder filter make the phase response of the ladder very weird, which is then counteracted by more highpass filters afterwards, and the result is hairy
but again, this whole signal path only has one special property (IMO) which is important in the 303 sound - the way the resonant filter lacks resonance on the bassy cutoff frequencies, unlike most other synths

the sound of the square wave is mostly in the very typical frequency-dependant waveshape
the waveshaper is not memoryless, there are two capacitors involved, and as a result, at low frequencies the square's middle edge has a soft knee, it gets sharper at higher frequencies, and the pulse-width also varies dramatically
the most audible aspect of the square is the frequency-dependant pulse-width

Id be lying if i said that i completely understood this, (no offense), but the part about the varying pulsewidth is interesting, especially since the 'driver-oscillator' really appears to be just a bog standard Saw. Is this actual research-data gathered from hardware-measurements or a hypothesis based on interpreting schematics? Id like to know more about this.

i spent a few years trying to synthesize the 303 (actually a x0xb0x) digitally
i've poked the circuit, burnth myself with the iron, and learned a few things about it

This is fun:

Well he just seems to be repeating what many other sources are saying.

He also says that the 303s Filter and Amp are controlled by an AR envelope, (he even draws an Attack/Release graph when he says it), so im not sure he can be considered a 'reliable source'.

yes, the guy who made that probably spent more time editing the video than preparing his facts
there is no A/R envelope.. there are two Decay envelopes, the one for the VCA has a fixed time, while the other one has a computer-controlled (switching) time based on the accent and a potentiometer.. it should also be noted that the VCA takes two envelopes.. the accent creates a third envelope (based on the main one) which is summed with the volume envelope and that gives the louder accent note effect

Heres another test setup. The signal goes Saw -> Waveshaper -> HPF, and sure enough im getting the exact same result again:

click to view animation


Interesting, isnt it. But dont get me wrong here, i really couldnt care less if im right or not. All im interested in is whats really going on, so if there is irrefutable proof that the 303s odd waveshapes are not a direct result of the highpassing i would absolutely like to see it.

your waveshaper is wrong, and you should forget about the highpass filters when looking at the oscillator waveform, because one highpass filter will not substitute for all the 10 highpass filters in the signal chain, no matter how much you tweak it
the actual waveform looks different.. you could check my link to the circuit simulator to see and play with it (because the waveform changes depending on the frequency)

here is an old video that shows it a bit.. that's the actual square wave of the x0xb0x

[DJ Warmonger]

Here you go, square-shaping circuit based on linked article (need Java to run):
http://www.falstad.com/circuit/circuitj ... 25+1+-1%0A

Thanks. I tried to check this out but the site doesnt open in my browser.

I take it its some kind of circuit simulator?

It's an interactive online circuit simulator called Circuit Simulator.

[ENV1]

you are not hearing the audible saw.

Im not sure what you mean here (language barrier?) but audible means 'what you hear', so of course you are hearing the audible Saw, in other words the Saw that comes out of the output.

you are heary an almost perfect saw, which goes thru about 10 highpass filters, a ladder filter, a VCA, and a mixer

Well, no. What comes out of the 303 is as far from a perfect Saw as it gets, in fact it doesnt even remotely look like one. Check this picture. These waveforms are from clips from a hardware 303. (Dips of Saw cut by filter.)

click for fullsize

the few highpass filters in the actual feedback path of the ladder filter make the phase response of the ladder very weird, which is then counteracted by more highpass filters afterwards, and the result is hairy
but again, this whole signal path only has one special property (IMO) which is important in the 303 sound - the way the resonant filter lacks resonance on the bassy cutoff frequencies, unlike most other synths

So what youre saying is that the highpassing is not responsible for the waveshapes?

If so then what is it in your opinion?

your waveshaper is wrong, and you should forget about the highpass filters when looking at the oscillator waveform, because one highpass filter will not substitute for all the 10 highpass filters in the signal chain, no matter how much you tweak it

Well, obviously this is just a simplified setup to demonstrate that the process of highpassing does have the effect of shaping the waveform into the very shape we are talking about. It may well be that there is more than just 1 HPF involved in the hardware, but thats not important, the point is that the process of highpassing does have that certain effect on the waveform, thus it stands to reason that this is what causes it in the hardware as well, simply because this bending effect occurs automatically (and inevitably) when youre running this half-saw/half-square kind of waveform through enough highpass, in fact it cannot give you any other shape. (And you get the Saw too when you narrow the pulsewidth enough.)

So keeping this in mind, why would Roland go to the trouble of installing a waveshaper that shapes a standard Saw into such unusual shapes when the highpassing automatically creates exactly these shapes anyway? Another question is; would these shapes, if generated pre-HPFs, even survive the highpassing or would they end up looking totally different? (Well, i guess that one is easy enough to find out.)

the actual waveform looks different.. you could check my link to the circuit simulator to see and play with it (because the waveform changes depending on the frequency)

here is an old video that shows it a bit.. that's the actual square wave of the x0xb0x

Well thats not a 303 Square. It may well be that the x0xb0x has one like that, but as you can see on the image above, the 303 has a completely different one. As for emulations: ReBirth has it too (although that one is not 100% correct either), ABL2 has it, and ABL3 Mode3 has it too.


EDIT: I just checked Venom and it has the same waveforms too. Did i misunderstand you? What exactly was the video supposed to show? The waveform before everything else?

[ENV1]

Here you go, square-shaping circuit based on linked article (need Java to run):
http://www.falstad.com/circuit/circuitj ... 25+1+-1%0A

Thanks. I tried to check this out but the site doesnt open in my browser.

I take it its some kind of circuit simulator?

It's an interactive online circuit simulator called Circuit Simulator.

How fitting. (The name i mean.)

Id love to check it out but unfortunately it wont open in my browser.

[subterfuge]

So keeping this in mind, why would Roland go to the trouble of installing a waveshaper that shapes a standard Saw into such unusual shapes when the highpassing automatically creates exactly these shapes anyway?

Not so fast. You are not proving anything comparing the audio output of a totally different softsynth to a theory you have about a specific part of an analog synth circuitry. Your original claim was that the commonly known order of saw to square via waveshaper is a myth. That was proven false before you ever made the claim and easily seen on schematics and the actual circuit.

Now you keep talking about the high pass filtering and apparently think that you can somehow define the usefulness of specific part of circuitry in one while comparing only the final audio output of both? Your scientific methods may need some reconsideration.

[ENV1]

Not so fast. You are not proving anything comparing the audio output of a totally different softsynth to a theory you have about a specific part of an analog synth circuitry.

Not sure what you mean, because i never compared any audio. I merely noticed that highpassing a certain waveform will give you the exact waveforms that come out of the 303 and thus wondered if this highpassing, not the waveshaper, is responsible for these unique waveshapes. Do you think that this is an unreasonable assumption?

Your original claim was that the commonly known order of saw to square via waveshaper is a myth. That was proven false before you ever made the claim and easily seen on schematics and the actual circuit.

I have not made any claims. I have submitted a hypothesis. Hypotheses are sometimes wrong and need to be adapted. This i have done as soon as i realized that a Saw oscillator is in fact way more likely. Is there more that i could or should have done in your opinion?

Now you keep talking about the high pass filtering and apparently think that you can somehow define the usefulness of specific part of circuitry in one while comparing only the final audio output of both?

Well, yes, i keep talking about the highpassing since this is what generates the effect with every other synth ive tried. And like i said, i wasnt comparing audio, i simply noticed that the 303 waveforms can be perfectly reproduced by highpassing a certain waveform, and now im trying to make sense of it. Wouldnt you agree that this is better than taking everything thats written in places like wikipedia at face value?

Your scientific methods may need some reconsideration.

May be so. Im sure every scientist knows the feeling.

[MadBrain]

you are not hearing the audible saw.

Im not sure what you mean here (language barrier?) but audible means 'what you hear', so of course you are hearing the audible Saw, in other words the Saw that comes out of the output.

What he means is that the output of the oscillator is basically a perfect saw. THEN this perfect saw gets potentially waveshaped into the square wave, and then that gets sent into the filter and amp, which contain a bunch of highpass filters, and that's what makes it look "bent".

you are heary an almost perfect saw, which goes thru about 10 highpass filters, a ladder filter, a VCA, and a mixer

Well, no. What comes out of the 303 is as far from a perfect Saw as it gets, in fact it doesnt even remotely look like one. Check this picture. These waveforms are from clips from a hardware 303. (Dips of Saw cut by filter.)

click for fullsize

When you put a perfect Saw through a filter (which contains HPFs), and an amplifier (which also contains HPFs), the result looks like that!

the few highpass filters in the actual feedback path of the ladder filter make the phase response of the ladder very weird, which is then counteracted by more highpass filters afterwards, and the result is hairy
but again, this whole signal path only has one special property (IMO) which is important in the 303 sound - the way the resonant filter lacks resonance on the bassy cutoff frequencies, unlike most other synths

So what youre saying is that the highpassing is not responsible for the waveshapes?

If so then what is it in your opinion?

He's saying that the filter (with HPFs in it) and the amplifier (with HPFs in it) are responsible for the waveshapes.

your waveshaper is wrong, and you should forget about the highpass filters when looking at the oscillator waveform, because one highpass filter will not substitute for all the 10 highpass filters in the signal chain, no matter how much you tweak it

Well, obviously this is just a simplified setup to demonstrate that the process of highpassing does have the effect of shaping the waveform into the very shape we are talking about. It may well be that there is more than just 1 HPF involved in the hardware, but thats not important, the point is that the process of highpassing does have that certain effect on the waveform, thus it stands to reason that this is what causes it in the hardware as well, simply because this bending effect occurs automatically (and inevitably) when youre running this half-saw/half-square kind of waveform through enough highpass, in fact it cannot give you any other shape. (And you get the Saw too when you narrow the pulsewidth enough.)

No, the fact that there are multiple HPFs in different parts of the signal chain IS important, because it changes how the filter behaves. The fact that you have HPFs inside the filter's resonant path lowers the resonance when the cutoff is low. You're NEVER going to get the same effect by adding HPFs before or after the filter (the resonance will stay the same) - the ONLY way of doing it is putting HPFs inside the filter.

So keeping this in mind, why would Roland go to the trouble of installing a waveshaper that shapes a standard Saw into such unusual shapes when the highpassing automatically creates exactly these shapes anyway? Another question is; would these shapes, if generated pre-HPFs, even survive the highpassing or would they end up looking totally different? (Well, i guess that one is easy enough to find out.)

What antto is saying is that there is only a waveshaper for producing the square wave. High pass filtering does NOT produce a square wave out of a saw wave no matter how hard you try. So obviously Roland had to put a waveshaper in the 303, to produce the square wave.

the actual waveform looks different.. you could check my link to the circuit simulator to see and play with it (because the waveform changes depending on the frequency)

here is an old video that shows it a bit.. that's the actual square wave of the x0xb0x

Well thats not a 303 Square. It may well be that the x0xb0x has one like that, but as you can see on the image above, the 303 has a completely different one. As for emulations: ReBirth has it too (although that one is not 100% correct either), ABL2 has it, and ABL3 Mode3 has it too.

Antto's video shows his emulation of the 303 square wave BEFORE IT GOES THROUGH THE FILTER (with HPFs in it) AND THE AMPLIFIER (with HPFs in it). Once you apply the filter and amp, the result will look like you expect.

EDIT: I just checked Venom and it has the same waveforms too. Did i misunderstand you? What exactly was the video supposed to show? The waveform before everything else?

Yes, the video shows the waveform before everything else.

[ENV1]

What he means is that the output of the oscillator is basically a perfect saw. THEN this perfect saw gets potentially waveshaped into the square wave, and then that gets sent into the filter and amp, which contain a bunch of highpass filters, and that's what makes it look "bent".

Well then we are in perfect agreement.

Because thats exactly what i was saying.

When you put a perfect Saw through a filter (which contains HPFs), and an amplifier (which also contains HPFs), the result looks like that!

Yes, a highpassed SawDown looks pretty much exactly right. (Not 100%, but close enough.)

This would indicate that the signal for the Saw is being tapped before the waveshaper, (therefore no width-modulation is required for the Saw to look right), and by extension that the assumption that the final shape is a direct result of the highpassing is correct.

No, the fact that there are multiple HPFs in different parts of the signal chain IS important, because it changes how the filter behaves. The fact that you have HPFs inside the filter's resonant path lowers the resonance when the cutoff is low. You're NEVER going to get the same effect by adding HPFs before or after the filter (the resonance will stay the same) - the ONLY way of doing it is putting HPFs inside the filter.

I meant its not important with respect to the bending of the wave, which as you can see requires no more than 1 HPF.

What antto is saying is that there is only a waveshaper for producing the square wave. High pass filtering does NOT produce a square wave out of a saw wave no matter how hard you try. So obviously Roland had to put a waveshaper in the 303, to produce the square wave.

Well thats exactly what i said.

I never said that highpassing a Saw creates a Squarewave, (which would be absurd), in fact the clip i posted earlier today shows very clearly that i have a waveshaper turning the Saw into the right shape.

Antto's video shows his emulation of the 303 square wave BEFORE IT GOES THROUGH THE FILTER (with HPFs in it) AND THE AMPLIFIER (with HPFs in it). Once you apply the filter and amp, the result will look like you expect.

The video shows the waveform before everything else.

Good. Thats all i wanted to know.

Thanks for taking the time to reply.

[antto]

the actual waveform looks different.. you could check my link to the circuit simulator to see and play with it (because the waveform changes depending on the frequency)

here is an old video that shows it a bit.. that's the actual square wave of the x0xb0x

Well thats not a 303 Square. It may well be that the x0xb0x has one like that, but as you can see on the image above, the 303 has a completely different one. As for emulations: ReBirth has it too (although that one is not 100% correct either), ABL2 has it, and ABL3 Mode3 has it too.

Antto's video shows his emulation of the 303 square wave BEFORE IT GOES THROUGH THE FILTER (with HPFs in it) AND THE AMPLIFIER (with HPFs in it). Once you apply the filter and amp, the result will look like you expect.

that's actually the real square wave from the x0xb0x.. it's not an emulation
that's the signal off of the waveform switch, fed into the soundcard thru a resistor, then the "scope" inverses the DC-blocking filters of the soundcard.. you can also hear the audio on the video (it's kinda quiet)

[Daru925]

In case you have massive, the Square-Saw II wavetable is pretty similar, with saw at 100%, use an allpass filter to get the inverted saw. Then have fun all sort or distortion and enveloppes^^

http://quadrophone.com/site/assets/file ... saw-ii.gif

[subterfuge]

Not sure what you mean, because i never compared any audio.

I said; "compared the audio output of a totally different softsynth to a theory you have about a specific part of an analog synth circuitry." No matter if you compared by listening, measuring or just looking at it, point was not how you did it but that you compared different things to begin with.

I have not made any claims. I have submitted a hypothesis. Hypotheses are sometimes wrong and need to be adapted. This i have done as soon as i realized that a Saw oscillator is in fact way more likely. Is there more that i could or should have done in your opinion?

I think you could've maybe said more clearly where you were wrong about your hypothesis like you presented it first, when you realized it. Yet after all the proof shown even in this topic you still say that "Saw oscillator is in fact more likely." There's no "more likely" to how the 303 oscillator is built, it is what it is. Your hypothesis may have changed through the topic to a more likely explanation of what you see/hear/measure, I'll give you that. But the facts of how the hardware was/is built don't change, that's actually why you had to adapt your hypothesis instead, right?

You asked what I think about high pass filtering vs. waveshaping but I honestly don't know how to answer because I still don't know what is your current stand and what exactly do you want me to compare? Others have given great explanations of how the real hw works, I have nothing new to add to that. Personally, I think every component in that board affects the whole, even down to single component types. I know that is too general opinion not really telling anything while hinting at probably too specific details to go into in this topic, but have to leave it at that.

Well, yes, i keep talking about the highpassing since this is what generates the effect with every other synth ive tried. And like i said, i wasnt comparing audio, i simply noticed that the 303 waveforms can be perfectly reproduced by highpassing a certain waveform, and now im trying to make sense of it. Wouldnt you agree that this is better than taking everything thats written in places like wikipedia at face value?

I'm not suggesting to take anything at face value. And yes, you did present a hypothesis, asking proof to contrary. There have been some excellent answers from real experts of this circuit like Antto, who not only knows the hardware aspect inside out but has also made pretty darn good software emulation of it. I appreciate how the discussion has been going and how there have been even some new things to think about this old and simple circuit, thanks for that everybody.

[antto]

Not sure what you mean, because i never compared any audio. I merely noticed that highpassing a certain waveform will give you the exact waveforms that come out of the 303 and thus wondered if this highpassing, not the waveshaper, is responsible for these unique waveshapes. Do you think that this is an unreasonable assumption?

you're missing the point, and i said that multiple times
the "303 waveform" especially the square wave, is not something static..
no amount of highpass filtering will give you a square wave like that
here's a picture.. that's captured from two notes, the second one is low pitched
(that's from the whole audio-output, not the bare oscillator)

that's really not a "square", not even a "pulse" ... it's a weird pulse with very specific characteristics, the most audible of which is the frequency-dependant pulse width.. we call it "303 square" for that reason

Well, yes, i keep talking about the highpassing since this is what generates the effect with every other synth ive tried. And like i said, i wasnt comparing audio, i simply noticed that the 303 waveforms can be perfectly reproduced by highpassing a certain waveform, and now im trying to make sense of it.

go ahead and highpass a sawtooth so it looks like on my picture
when you do, just change the source waveform to a "square" without touching the highpass filters.. you can waste very long.. long time.. and you won't get a 303 square if your source square doesn't have the right characters, and pretty much nothing has, unless it's an intentional emulation of the 303 square.. it's just a weird pulse

[NAD]

I don't really have much to contribute, just thought I'd throw this up here. This shows the waves from the latest AudioRealism Bass Line 3 for each of three VCO models on square and saw. According to the manual #2 is meant to be the most authentic emulation. To me #3 seems more similar to the other waveform images we've seen in this thread so far.

Untitled-1.png

[ENV1]

I said; "compared the audio output of a totally different softsynth to a theory you have about a specific part of an analog synth circuitry." No matter if you compared by listening, measuring or just looking at it, point was not how you did it but that you compared different things to begin with.

Sorry, but i dont see your point. (Seriously.) Nor can i see why it would constitute a problem even if i had been doing comparisons of any kind, which as i said was not the case. I ran into this phenomenon purely by accident, and recognized it for what it was only because i happened to know how the 303 Square looks. In other words there was no planning involved, it was simply something i happened to notice while i was doing something else.

I think you could've maybe said more clearly where you were wrong about your hypothesis like you presented it first, when you realized it.

Noted. Personally i dont see how i could have made it any clearer, but noted.

Yet after all the proof shown even in this topic you still say that "Saw oscillator is in fact more likely."

No, actually this is not what i still say. When i said "This i have done as soon as i realized that a Saw oscillator is in fact way more likely" i was obviously referring to the time when i began to suspect that my initial assumption might be incorrect, but the thread has progressed since. I have not questioned the oscillator type since my 3rd post in this thread.

Your hypothesis may have changed through the topic to a more likely explanation of what you see/hear/measure, I'll give you that. But the facts of how the hardware was/is built don't change, that's actually why you had to adapt your hypothesis instead, right?

Well...yeah. But then again thats not surprising? I mean if i had been certain about everything already, what would be the point in asking questions?

You asked what I think about high pass filtering vs. waveshaping but I honestly don't know how to answer because I still don't know what is your current stand and what exactly do you want me to compare?

There must be some sort of misunderstanding. I have directed only one question at you personally and that was on page 1.