Hot DCO Discussion!

Discussions about anything analog, digital, MIDI, synth technology, techniques, theories and more.
User avatar
cbjlietuva
Junior Member
Junior Member
Posts: 102
Joined: Thu Nov 25, 2004 10:42 pm
Gear: Roland Juno 6 · Rhodes 73 Suitcase · Fender '63 jBass · Fender Strat+ · Casio SK1
Location: A Honda Civic down by the foothils

Post by cbjlietuva » Sat Sep 08, 2007 1:55 am

otto wrote:
JUGEL wrote:I'm a bit fuzzy on this stuff. Where would something like my Akai AX-60 come into play? Hybrid of some kind?
I thought an AX-60 was VCO, no?
it is- it's just a CEM chip, an IC configuration, similar to the VCOs found in the Prophet 5... no clock, just voltage.
I've seen things you people wouldn't believe...Star Wars on Laserdisc

User avatar
Automatic Gainsay
Synth Explorer
Synth Explorer
Posts: 3962
Joined: Wed Feb 14, 2007 12:22 am
Real name: Marc Doty
Gear: Minimoog, 2600, CS-15, CS-50, MiniBrute, MicroBrute, S2, Korg MS-20 Mini, 3 Volcas, Pro 2, Leipzig, Pianet T, Wurli 7300, Wurli 145-A, ASR-10, e6400.
Band: Godfrey's Cordial
Location: Tacoma
Contact:

Post by Automatic Gainsay » Sat Sep 08, 2007 2:03 am

steveman wrote:People are confusing themselves further by talking about how the control signals are generated, this is actually irrelevant as to whether an oscillator is analogue or digital. All polysynths (and some monos eg. Pro-One) generate CVs for the VCOs and VCFs from a digital keyboard scanning system. (Pedants - please don't bring up divide-down synths here... ;) )
But what about divid- oh, wait, nevermind. :)

Thank you for your post... it is very clear and exceedingly helpful.
‎"I am not able rightly to apprehend the kind of confusion of ideas that could provoke such a question." -Charles Babbage
"Unity and Mediocrity are forever in bed together." -Zane W.
http://www.youtube.com/automaticgainsay

User avatar
OriginalJambo
Synth Explorer
Synth Explorer
Posts: 2560
Joined: Sun Jul 09, 2006 12:04 am
Gear: Check my sig
Location: Scotland, United Kingdom

Post by OriginalJambo » Sat Sep 08, 2007 5:24 pm

Technicalities aside, this my 2p:

VCO - completely analogue, pitch varies, waveshape varies, not as stable and affected by temperature. Zero beat not possible between two oscillators.

DCO - completely analogue audio signal but pitch digitally controlled, waveshape varies, very stable and not affected by temperature. Does not suffer from issues related to digital to analogue conversion - e.g. aliasing.

DGO - completely digital, a mathematical representation of the sound, uses DSP, fed through D/A converter so we can hear it, does not drift at all, perfect stability but suffers from aliasing and maybe other artefacts.

Is that good enough for the basics?
Last edited by OriginalJambo on Sat Sep 08, 2007 7:05 pm, edited 1 time in total.

User avatar
micahjonhughes
Active Member
Active Member
Posts: 703
Joined: Fri Mar 09, 2007 4:27 pm
Location: Chicago

Post by micahjonhughes » Sat Sep 08, 2007 6:04 pm

I think that we are starting to confuse control voltages with some other aspect of DCOs. The control voltage can be perfect and digitally controlled as in a Prophet 5 but the synth and its oscilators are completely analog. And even if the control voltage is perfect, it does not mean that the oscillator puts out a perfect wave with constant frequency.

DCOs seem to need a digital clock that tells when the cycle of the waveform should begin. This way each cycle is of the same exact same length or frequency as the last.

I'm interested in various ways this technology was implemented and hope that we can keep that discussion going.

User avatar
nathanscribe
VSE Review Contributor
VSE Review Contributor
Posts: 2889
Joined: Fri Aug 11, 2006 1:03 pm
Location: The right side of the Pennines
Contact:

Post by nathanscribe » Sat Sep 08, 2007 8:11 pm

memo wrote:I'm looking at the Juno 60 Service Notes right now
Not far off, but there's no 'sawtooth generator IC' - the key CV is multiplexed to an integrator (opamp, cap and transistor) which is clocked on or off by the timer pulse. The waveform is analogue all the way, but the timing signal is generated by a clock.

I wonder how many people on VSE actually know what 'digital' and 'analogue' actually mean (no directed at you, memo - just a general comment). My modular's LFO puts out a square wave at 5v, but it's not exactly a digital signal... there's a big difference between a rising voltage that looks like a sawtooth and a train of pulses flowing at high speed that, in binary, reperesent the successive ascending levels of a sawtooth and are converted into analogue voltages at the end.

Heads, bricks, walls.... I love the internet.

User avatar
memo
Junior Member
Junior Member
Posts: 173
Joined: Mon May 29, 2006 3:53 am
Location: Berkeley, CA
Contact:

Post by memo » Sat Sep 08, 2007 8:43 pm

nathanscribe wrote:Not far off, but there's no 'sawtooth generator IC' - the key CV is multiplexed to an integrator (opamp, cap and transistor) which is clocked on or off by the timer pulse. The waveform is analogue all the way, but the timing signal is generated by a clock.
I kind of figured IC16 was doing the major work on the wave generation, timed by the pulses. Either way, it's definitely analog - only clocked digitally.
I wonder how many people on VSE actually know what 'digital' and 'analogue' actually mean
I've always thought of it like this:

An analog signal is analogous to its purpose. You could hack into any point of an analog synth and route it into another analog synth, and providing that you matched impedance and signal levels, it would actually work. You could create a modular Pro-One that receives FM from a Juno 60, if you so desired, since each and every signal in it isn't merely representing a signal's purpose - it IS that signal.

Compare that to a Korg Triton - if you were somehow able to hack into the signal between the oscillator DSP and the filter DSP, you couldn't route that signal into another synth, or hear the raw oscillator sound. It would just be jumbled data to anything that received the signal - it would be like putting an data CD into an old audio CD player - it would sound like jumbled noise.

The signal is merely representing a raw oscillator - the 1's and 0's that make up a digital signal are not analogous to their purpose.
Image

User avatar
memo
Junior Member
Junior Member
Posts: 173
Joined: Mon May 29, 2006 3:53 am
Location: Berkeley, CA
Contact:

Post by memo » Sat Sep 08, 2007 8:54 pm

OriginalJambo wrote:Technicalities aside, this my 2p:VCO DCO DGO
I think there are four, since each sounds distinctly different, and there is an order from warmth/instability to sterile/stability.

VCO: Voltage Controlled Oscillator (Classic 70's synths)

DTO: Digitally Tuned (Analog) Oscillator (Early 80 Polys, Andromeda)

DCO: Digitally Controlled (Analog) Oscillator (Junos, PEK)

DGO: Digitally Generated Oscillator (Softsynths, Digital Synths)

A Hybrid is a DGO going through a VCF/VCA
Image

User avatar
OriginalJambo
Synth Explorer
Synth Explorer
Posts: 2560
Joined: Sun Jul 09, 2006 12:04 am
Gear: Check my sig
Location: Scotland, United Kingdom

Post by OriginalJambo » Sat Sep 08, 2007 10:13 pm

memo wrote:I think there are four, since each sounds distinctly different, and there is an order from warmth/instability to sterile/stability.
I see what you are saying with the DTO but to me there are three major oscillator types. Ones that drift, ones that don't and ones that are essentially just computer code. ;)

A DTO is a VCO that has voltage regulation as opposed to being digitally clocked, yeah?
A Hybrid is a DGO going through a VCF/VCA
I agree wholeheartedly. The ESQ-1 and DW-8000 are hybrids and the Juno and JX are not to me.

As for the differences between analogue and digital, I think of it like this:

- An analogue voltage can be anything between 0 and x volts. The number of possible values is infinite.

- A "digital voltage" (there isn't even such a thing!) is simply where you decide on two states representing 1 (e.g. +5V) and 0 (e.g. 0V), on and off etc. with no possible values in between.

MitchK1989
Active Member
Active Member
Posts: 663
Joined: Mon Apr 24, 2006 1:46 am
Location: Nova Scotia

Post by MitchK1989 » Sat Sep 08, 2007 10:33 pm

OriginalJambo wrote:- A "digital voltage" (there isn't even such a thing!) is simply where you decide on two states representing 1 (e.g. +5V) and 0 (e.g. 0V), on and off etc. with no possible values in between.
You say that like its a bad thing, without also noting that because the bandwidth is so much higher in a digital system, you can generate an absurdly fine resolution and recreate all the middle values just fine using say 1000 digits of 0/1 instead of just 1 sample of X. And as you can check for yourself on a calculator, you can still write 435 in binary, it just comes out as 110110011...

Really, the sample rate and bit rate of modern digital systems allows for all but the highest pitches without aliasing when implemented well, and any digital synth that sounds bad sounds bad due to poor design/coding, not due to any necessary limitations of digital.

User avatar
OriginalJambo
Synth Explorer
Synth Explorer
Posts: 2560
Joined: Sun Jul 09, 2006 12:04 am
Gear: Check my sig
Location: Scotland, United Kingdom

Post by OriginalJambo » Sat Sep 08, 2007 10:53 pm

MitchK1989 wrote:You say that like its a bad thing, without also noting that because the bandwidth is so much higher in a digital system, you can generate an absurdly fine resolution and recreate all the middle values just fine using say 1000 digits of 0/1 instead of just 1 sample of X. And as you can check for yourself on a calculator, you can still write 435 in binary, it just comes out as 110110011...
I didn't say it was a bad thing at all. That's the basic principal of digital - 2 possible states.

Of course anything can be accurately represented with binary, and therefore digitally, given the right number of bits however at a very basic level that's all digital is...0s and 1s.
Really, the sample rate and bit rate of modern digital systems allows for all but the highest pitches without aliasing when implemented well, and any digital synth that sounds bad sounds bad due to poor design/coding, not due to any necessary limitations of digital.
I totally agree, but you still don't get these artefacts at all in a fully analogue signal path. Just pointing that out.

On the other hand, one huge advantage to fully digital synths is that there is no noise generated by them unless they are converted back to the analogue domain. The onboard distortion effect will never increase the noise floor because there isn't one to begin with.

User avatar
memo
Junior Member
Junior Member
Posts: 173
Joined: Mon May 29, 2006 3:53 am
Location: Berkeley, CA
Contact:

Post by memo » Sun Sep 09, 2007 3:51 am

OriginalJambo wrote:I see what you are saying with the DTO but to me there are three major oscillator types. Ones that drift, ones that don't and ones that are essentially just computer code. ;)

A DTO is a VCO that has voltage regulation as opposed to being digitally clocked, yeah?
A DTO is an analog synth that receives regular background digital tuning - they're not clocked like a DCO, but they're constantly corrected so they never shift that much. It's one of the reasons why 80s polys are different from 70s polys.

They're kind of in between VCO and DCO in both warmth and stability.

Of course, it's not an official term at all - it's just me being a bit too picky on trying to categorize.

I think DGO should be an official term, since people constantly confuse them and DCO synths.
Image

User avatar
elmosexwhistle
VSE Review Contributor
VSE Review Contributor
Posts: 283
Joined: Thu Nov 17, 2005 11:27 am
Real name: Alex
Band: Elmo Sexwhistle
Location: Birmingham
Contact:

...

Post by elmosexwhistle » Sun Sep 09, 2007 5:16 am

i kinda thought the filter was more important than whether the square/saw/tri/pulse waves are created digitally or not...but i agree DCO seems to be too broad a term...x

bitrex
Newbie
Newbie
Posts: 3
Joined: Thu Aug 09, 2007 4:22 am

Post by bitrex » Sun Sep 09, 2007 6:23 am

I have experience working on the Roland Juno 106 and the Alpha Juno 2, so if anyone's interested I'll explain how the Juno 106's implementation of a DCO works (it's quite similar to the first part of solderguy's explanation of the Korg synth, and the Alpha Juno is similar as well.)

The 106 contains a master crystal oscillator that creates an 8 MHz square wave. This is first sent to a prescaler IC to bring the oscillator down closer to the audio range, and then to two 82C53 programmable timers. Each chip is responsible for generating the frequency of 3 voices, hence an 6 voice synth.

When a key is pressed, or the 106 receives a MIDI note on message, the microprocessor compares the MIDI note data to a lookup table in ROM, sees how many voices are in use, then activates the appropriate 82C53 channel and sends an 8 bit message over one of the address/data buses. This 8 bit message tells the programmable timer by what factor to divide down the master oscillator to get the frequencies of the note(s) that chip is responsible for. The frequencies aren't the exact intervals of the equal tempered scale (each semitone the 12th root of 2 above the previous), but because the master clock is so high, they are very close.

The square waves then go on to the 3 waveshaping ICs, MC5534s, each of which is responsible for 2 voices and is on the 3 voice boards. The output square wave is basically applied to an internal op amp that provides PWM, and the sub oscillator is provided by just a flip flop that divides the incoming square wave by 2. The sawtooth is generated by a constant current source charging a capacitor, which is discharged through a transistor switch turned on and off by the incoming clock. Unfortunately, as the frequency increases higher up the keyboard the charging time with respect to the frequency becomes less and less, so a proportional control voltage is applied to the constant current source to increase the charing current and keep the sawtooth output level steady. The selected waveforms are then summed up and sent out to the VCF/VCA chip.

The 106 contains a single DAC, which is actually not even a single chip but is constructed from a resistor array and other discrete components! I believe this is because it has 14 or so bits of resolution, and back in the early 80s single chip DACs with that kind of resolution probably cost several hundred dollars. The DAC creates analog voltages to control all the oscillators, VCAs, and VCFs through a complicated multiplexing scheme involving 4051 ICs and op-amp sample and holds. It's amazing how all the things the synth does appear to happen at the same time, but really the microprocessor is furiously sending data through that one DAC and turning on and off all those IC switches to keep the control voltages updated.

So, I guess to sum up, are the oscillators "digital?" Not in the way I think of digital nowadays, i.e. the AUDIO signal from a synth like the 106 is not coming out of a digital to analog converter. The sawtooth generator is similar to a VCO sawtooth generator, it's just locked to a steady master pulse, instead of being free-running as it is in some other synths. The master pulse IS under digital control though, as its frequency is controlled by the divider chip which is controlled by the microprocessor. I think the term "Digitally Controlled Oscillator" should just refer to synths like the 106 that use this scheme, and modern synths like the Virus etc. that are essentially computers hooked to D/As could be referred to as having just "Digital Oscillators" or just the popular "Virtual Analog."

So there ya go. Did I end the debate? :P

User avatar
i_watch_stars
Active Member
Active Member
Posts: 271
Joined: Mon Jun 26, 2006 9:49 am
Location: Seattle

Post by i_watch_stars » Sun Sep 09, 2007 11:11 am

bitrex wrote:I have experience working on the Roland Juno 106 and the Alpha Juno 2, so if anyone's interested I'll explain how the Juno 106's implementation of a DCO works (it's quite similar to the first part of solderguy's explanation of the Korg synth, and the Alpha Juno is similar as well.)

The 106 contains a master crystal oscillator that creates an 8 MHz square wave. This is first sent to a prescaler IC to bring the oscillator down closer to the audio range, and then to two 82C53 programmable timers. Each chip is responsible for generating the frequency of 3 voices, hence an 6 voice synth.

When a key is pressed, or the 106 receives a MIDI note on message, the microprocessor compares the MIDI note data to a lookup table in ROM, sees how many voices are in use, then activates the appropriate 82C53 channel and sends an 8 bit message over one of the address/data buses. This 8 bit message tells the programmable timer by what factor to divide down the master oscillator to get the frequencies of the note(s) that chip is responsible for. The frequencies aren't the exact intervals of the equal tempered scale (each semitone the 12th root of 2 above the previous), but because the master clock is so high, they are very close.

The square waves then go on to the 3 waveshaping ICs, MC5534s, each of which is responsible for 2 voices and is on the 3 voice boards. The output square wave is basically applied to an internal op amp that provides PWM, and the sub oscillator is provided by just a flip flop that divides the incoming square wave by 2. The sawtooth is generated by a constant current source charging a capacitor, which is discharged through a transistor switch turned on and off by the incoming clock. Unfortunately, as the frequency increases higher up the keyboard the charging time with respect to the frequency becomes less and less, so a proportional control voltage is applied to the constant current source to increase the charing current and keep the sawtooth output level steady. The selected waveforms are then summed up and sent out to the VCF/VCA chip.

The 106 contains a single DAC, which is actually not even a single chip but is constructed from a resistor array and other discrete components! I believe this is because it has 14 or so bits of resolution, and back in the early 80s single chip DACs with that kind of resolution probably cost several hundred dollars. The DAC creates analog voltages to control all the oscillators, VCAs, and VCFs through a complicated multiplexing scheme involving 4051 ICs and op-amp sample and holds. It's amazing how all the things the synth does appear to happen at the same time, but really the microprocessor is furiously sending data through that one DAC and turning on and off all those IC switches to keep the control voltages updated.

So, I guess to sum up, are the oscillators "digital?" Not in the way I think of digital nowadays, i.e. the AUDIO signal from a synth like the 106 is not coming out of a digital to analog converter. The sawtooth generator is similar to a VCO sawtooth generator, it's just locked to a steady master pulse, instead of being free-running as it is in some other synths. The master pulse IS under digital control though, as its frequency is controlled by the divider chip which is controlled by the microprocessor. I think the term "Digitally Controlled Oscillator" should just refer to synths like the 106 that use this scheme, and modern synths like the Virus etc. that are essentially computers hooked to D/As could be referred to as having just "Digital Oscillators" or just the popular "Virtual Analog."

So there ya go. Did I end the debate? :P
Great informative post!

I'm wondering though...which has more "difference" in your opinion; the difference between a VA and a DCO, or the diff between a DCO and a VCO?

I have all three types of synths and I can really only tell the difference in the VCO vs DCO catagory...the VA and DCO raw oscillators sound the same to me, its the analog filters in that case which seperates them.

User avatar
nathanscribe
VSE Review Contributor
VSE Review Contributor
Posts: 2889
Joined: Fri Aug 11, 2006 1:03 pm
Location: The right side of the Pennines
Contact:

Post by nathanscribe » Sun Sep 09, 2007 12:27 pm

Excellent post, bitrex. :)

Post Reply