Tetra... Prophet 08 rack or something else?

[Dogboy73]

Nicely put Mr Bastard

Indeed. It's the way it is. Your never going to get a vintage synth enthusiast admitting that any modern equivelent is as good.

I'm no "vintage enthusiast" and I resent the implication.

I wasn't implying that you are one such enthusiast, ghetto. But also I wasn't having a pop at such people. I don't blame them for their views on modern equivelents of vintage machines. For such people the vintage synths offer something over the modern day equivelents that is not always quantifiable. Owners of Mini Moog Model D's will say that a Moog Voyager just doesn't cut it even though, for all intents & purposes, the Voyager is the modern day equivelent of the Model D. Personally I'd rather have the more reliable modern machine with nice new parts & a warranty rather than something that's as old as I am.

This entire "DSI is all analog" is just marketing nonsense.

True. There's a definite buzz around the whole vintage/analogue synth thing that companies use to promote their products. I think Roland are one of the worst offenders here though. Using names of classic machines on their modern range in order to promote them. I am off course talking about the SH-201 & Juno G. I don't think these machines should carry the names of the classics before them as they have little in common with those machines apart from the name! Roland use them purely because of the reputation these machines have. Now imagine if Roland decided to re-launch a proper SH-101 or Juno synth

[fac]

At minimum an analog synth has analog oscillators.

The Prophet 08 oscillators *are* analog.

No, they're not.

Yes, they are. A pulse signal from a clock generator/divider is analog since it is a continuous voltage.

I don't know who told you that but it's not correct. It is a digital signal, either 0 or V++ volts. It cannot be V++/2 or V++/k for any k.

I understand where you're coming from, but in that case, a square wave from an analog oscillator would also be digital, since it can be only -5 or +5 V (or whatever amplitude you'd like).

But that is not correct. Because both the analog square wave and the clock signal are actually continuous signals. The voltage does not magically jump from 0 to 5V... there is a very short, but continuous transition, so it is completely possible to observe, for example, a voltage of 1V or 2.3455V in a clock signal. It does not matter how it is produced or what it will be used for, a clock signal is as analog as an analog square wave.

The "digital"-ness of a signal lies in its representation. A digital signal, by definition, is discretized in both time and voltage. It is a signal that is represented as a series of numbers with a finite precision, and thus is subject to aliasing and other artifacts. In the case of the Prophet 08, and many other DCO-based synths (Matrix 6, Juno-106, etc), the audio signals are never represented in this way. There are no ADC's or DAC's in the audio signal path, so the signal never becomes digital, and as a result, there is no aliasing.

A DCO (Digitally Controlled Oscillator) outputs an analog waveform which starts as an analog clock pulse and uses integrators and waveshaping to convert the pulse signal into a sawtooth, triangle, etc. The initial pulse signal is obtained by a frequency divider applied to a high-frequency clock. A frequency divider is basically a modulo-N digital counter, and that's why these devices are called "digitally controlled oscillators", but the output signal is completely analog because it is never, at any stage, represented as a stream of finite precision numbers; it is never sampled or quantized. if the signal were digital at any point, it would require a DAC to convert it back to the analog domain. I built a DCO-based monosynth once, and trust me, I didn't use any ADCs or DACs and I was always able to measure the signal with an analog oscilloscope at any point in the path.

I'm not saying with this that "DSI is all analog"... it definitely isn't. But it *is* analog where it matters (audio path), and digital where it's useful (control path). No, it doesn't sound exactly like a vintage VCO because the DCO's are much more stable (to be honest, the "slop" parameter doesn't do much to add inestability), but that doesn't make the sound less analog. It sounds much better than any VA I've used for those kind of sounds and is much more flexible modulation-wise than most vintage polysynths. The only real downside of DSI synths is that Dave chose to use some crappy encoders, but apparently, they're being changed for better ones and/or potentiometers.

[ghettosynth]

...

[Dogboy73]

We've had software V's hardware, analogue V's digital & now we're onto analogue V's analogue! This is getting silly now. As far as I'm concerned a DCO is the same as a VCO except with the added bonus of digital tuning & thus doesn't have the tendency to drift out of tune like a VCO! For me this is an improvement over the VCO because you have something that is better behaved & can be controlled to behave like a VCO anyway. The Prophet 08 & the Tetra have the 'osc slop' control for this & it works a treat for adding the kind of subtle movement you get with VCO's. Then there's always other modulation sources that can do the same thing. So you have something that sounds just as good, is better behaved & more versatile. Ghetto mentioned that there have never been that many true analogue polysynths that were 100% analogue. There's a reason for this - they were crap! Badly behaved behomoths that wouldn't stay in tune if there lives depended on it! The Alesis Andromeda was an attempt to rectify the problems associated with this kind of thing in a modern synth. It failed. And so ends the balad of the analogue polysynth. R.I.P.

[ghettosynth]

...

[ghettosynth]

...

[zerocrossing]

Let me summarize what I've gathered by all of this chit chat...

Polysynths with VCOs... rare and very expensive. Old ones are problematic, a new one like the Andromeda or Omega 8 will set you back $3K. Nice if you're wealthy or a professional musician.

Vintage Polysynths with DCOs. Often "meh." I've owned a Juno 106 and I just spent a lot of time listening to MSK70 demos and they sound pretty good. Not great. Can be had fairly cheap, but I'm not sure they're significantly better than some of my better software emulations to be worth having a chunk of hardware taking up space in my world.

Modern Polysynths with DCOs. Other than DSI there any choice?

And again... I happen to really dig the DSI stuff. I can't figure out why people wouldn't be. Sometimes it's about good enough because if you wait for perfection you may end up dead before it strikes.

ghettosynth, what synths do you own and why did you pick them? I'd love to know.

[fac]

As I've said all along, it's a hybrid, digitally controlled oscillators are not the same thing as analog oscillators. Until now, nobody has ever tried to make that claim.

What is often referred to as a "VCO Core" is the part of a VCO that generates the timing pulse. The "core" of a DCO is a digitally generated timing pulse.

In simple terms, the heart of a DCO is digital, not analog.

Yes, the core of a DCO is a pulse signal which is generated by a frequency divider, which is a digital component. What I'm saying, however, is that the signal itself is fully analog. It is perfectly possible for a digital device to generate an analog signal, because in fact, all digital circuits are made from analog components.

The actual difference between an analog and a digital signal does not lie on the circuitry used to produce it. It lies in its representation. An analog voltage signal is continuous both in time and amplitude. A digital signal is discrete in both time and amplitude. With a digital signal, you know its amplitude at times 1, 2, 3, etc, but not at time 1.2 or 3.58. Also, the values of a digital signal are represented with finite precision... there's only a finite set of values you can choose from. An analog signal (not circuitry, a signal) has a defined value at each instant, and that value belongs to an infinite real interval.

This is not a "belief". These are standard definitions, which you can find on any signal processing book. Under these definitions, the signal output by a DCO *is* analog. It never enters the digital domain, not even when it starts as a clock pulse. Yes, the pulse is obtained from a frequency divider which uses a digital counter, but it is the counter that is digital (theoretically represented with a series of numbers), not the clock signal. If the clock signal were digital, one would have to be able to determine its sampling rate and finite precision (e.g., how many bits are used to represent each sample), and it would be also subject to aliasing. But it isn't.

Under these definitions, DSI is not lying to anyone. Saying that their synths have an "all analog signal path" is quite more accurate than saying their synths are "all analog", and to my knowledge, Dave has never stated the latter.

Similarly, saying "a DCO is analog" is very different from "a DCO outputs a 100% analog signal". The former is false, since a DCO is clearly a hybrid design, but the latter, according to common definitions, still holds. Though I'm not sure if you'd agree with this.

[zerocrossing]

As I've said all along, it's a hybrid, digitally controlled oscillators are not the same thing as analog oscillators. Until now, nobody has ever tried to make that claim.

What is often referred to as a "VCO Core" is the part of a VCO that generates the timing pulse. The "core" of a DCO is a digitally generated timing pulse.

In simple terms, the heart of a DCO is digital, not analog.

Yes, the core of a DCO is a pulse signal which is generated by a frequency divider, which is a digital component. What I'm saying, however, is that the signal itself is fully analog. It is perfectly possible for a digital device to generate an analog signal, because in fact, all digital circuits are made from analog components.

The actual difference between an analog and a digital signal does not lie on the circuitry used to produce it. It lies in its representation. An analog voltage signal is continuous both in time and amplitude. A digital signal is discrete in both time and amplitude. With a digital signal, you know its amplitude at times 1, 2, 3, etc, but not at time 1.2 or 3.58. Also, the values of a digital signal are represented with finite precision... there's only a finite set of values you can choose from. An analog signal (not circuitry, a signal) has a defined value at each instant, and that value belongs to an infinite real interval.

This is not a "belief". These are standard definitions, which you can find on any signal processing book. Under these definitions, the signal output by a DCO *is* analog. It never enters the digital domain, not even when it starts as a clock pulse. Yes, the pulse is obtained from a frequency divider which uses a digital counter, but it is the counter that is digital (theoretically represented with a series of numbers), not the clock signal. If the clock signal were digital, one would have to be able to determine its sampling rate and finite precision (e.g., how many bits are used to represent each sample), and it would be also subject to aliasing. But it isn't.

Under these definitions, DSI is not lying to anyone. Saying that their synths have an "all analog signal path" is quite more accurate than saying their synths are "all analog", and to my knowledge, Dave has never stated the latter.

Similarly, saying "a DCO is analog" is very different from "a DCO outputs a 100% analog signal". The former is false, since a DCO is clearly a hybrid design, but the latter, according to common definitions, still holds. Though I'm not sure if you'd agree with this.

I get up each morning to a digital alarm clock, yet I'm pretty sure I'm analog, though ghettosynth would disagree with that too.

[Shy]

I get up each morning to a digital alarm clock, yet I'm pretty sure I'm analog

Bah, mechanical is better.

[ghettosynth]

...

[ghettosynth]

...

[ghettosynth]

...

[fac]

From wikipedia:

In computer architecture and other digital systems, a waveform that switches between two voltage levels representing the two states of a Boolean value (0 and 1) is referred to as a digital signal, even though it is an analog voltage waveform, since it is interpreted in terms of only two levels.

Yeah. Thanks for the reference. Specifically the part where it affirms that the signal is actually analog, and that it is only referred as a digital signal in the context of computer architecture and digital systems.

Moreover, the same wikipedia page states that:

Although in a highly simplified and idealised model of a digital circuit we may wish for these transitions to occur instantaneously, no real world circuit is purely resistive and therefore no circuit can instantly change voltage levels. This means that during a short, finite transition time the output may not properly reflect the input, and indeed may not correspond to either a logically high or low voltage.

which further indicates that one may observe values different than 0 or 1 in the signal. Moreover, how does a pulse signal like this is distinguished from an analog pulse wave (which also alternates between low and high states). You do know you can, for example, clock a sequencer with a VCO's pulse wave, right? Does that make the signal less analog?

In any case, once the pulse signal goes through the integrator/waveshaper, it no longer represents a boolean value. How do you call that, then?

Furthermore, the wikipedia page also makes a distinction between two possible definitions of digital signal; the other one, which you keep ignoring, is:

In conceptual summary, a digital signal is a quantized discrete-time signal; a discrete-time signal is a sampled analog signal.

The output of a DCO does not, at any stage, fulfill this definition, either.

[ghettosynth]

...