O.K.. I just had a magical 40 minutes! Although I have been working with synths for decades and also started to repair and convert them years ago I still never had a moment when everything totally fell into place. Now that moment has come. Hurray, I at last understand how oscillators work on a component level! Suddenly all earlier, rather random insights fall into a linear tale. So from now on it will therefore be much easier to juggle all that information around. It's comparable to when somebody first made all the bits about music theory come together for me. I can still remember that magical moment as I will remember this one. So thank you very, very, very much!
Yeah! Sorry if that sounded pompous but it is the actual truth. That is the nice side of the internet. Someone from the other side of the world (I'm Dutch!) can really make a difference because he simply has the inclination to share his knowledge without a preconceived plan for personal (economic) gain. So I'm affraid I'll have to repeat it again: THANKS!
I add my voice to Marc. With decades of playing and fixing synths, I thought I was understanding DCO, but your video confirmed I was not. You lighted up all the black spots I had. Many thanks for that. However, the dark side of this video is .... we want more! ;-)
Great video, really enjoyed watching that. You kinda guessed right - the DeepMind12 has 100MHz master oscillator and is divided down to reach low pitch using a combination of 32 bit and 16 bit timers, plus we're using higher res DACs so don't need the range switch for the DCO CV.
Rob Belcham ha fancy to see your comment here. This was a great video and nice to see you chiming in so I don't have to twist your ear about it in chat lol! Such a great video
I'm such a fan of your channel! One of the most informative, comprehensive synth repair channels anywhere. Thanks for taking time to share and educate us synth lovers. I have a friend's Juno 106 in my studio that I'm trying to repair and the poor thing is pretty bad. Wish me luck, your videos really help!
Wow! Thank you for this concise and thorough explanation of the master clocks! I've always loved the sound of the Poly61, and felt that the 106 sounded a little too "tame". Now I fully understand why!
This video was wonderful. You did such an excellent job explaining all of this. If you're not a teacher, I think you would have been a fantastic one. Thank you so much for making this video.
Thanks for a very interesting and informative lesson! I really enjoyed this video, and - oops - just realized that I watched it all with the sound going through the Elektron Analog Heat sound processor set to the "Saturation" tape-like gentle distortion circuit. I love the oldschool tape recorder instruction movie feeling! Cheers! ;-)
wonderful tour of these schematics, thank you for taking the time to make this. a few points stand out that are worth noting. 1: thinking about a pnp as "normally closed" can get you into trouble. BJTs are all normally open! 2: in both cases of rising and falling saws, the rate is set by the current into the integrator circuit (op amp with cap out to negative input). the resistor in series with the transistor sets the discharge rate and limits the current through the transistor. 3: keeping the sawtooth level constant at different frequencies is the job of the microcontroller, and inherently tied to the ramp rate. this is done with the dacs. 3: rising and falling sawtooths sound identical, this is a fundamental truth of waveforms and harmonics. 4: Poly 800 doesn't have DCOs like this, but instead builds all of its waves by summing pulses!
Ha that was great. I never considered the impact of using a voltage controller osc as a clock for a DCO. Very cool and thanks for sharing this knowledge. Wow, and then you can modulate this clock source. Eye opening!
Excellent and timeless video! I was sent here by JX3D on the ranzee discord. Even with a good level of understanding electronics, I think you have just doubled my understanding of synth circuits in the space of one video. Thank you for 'knowledge bomb' (even if I am 6 years late).
You are the best for us synth geeks who are also our own techs; those of us who really want the details and some great ideas for mods. Thank you for showing us so much. Frank S
Amazing. Simply amazing, sir! I do look forward to your video uploads, as they are all so good, and not only interesting, but very educational. Like so many others following, I now can appreciate oscillator circuits better - I can't thank you enough!
Kaleb, I can't believe I watched this whole thing! This was So interesting and well presented. Thx for spending the hours to put this together and sharing.
Superb educational video! Thank you so much for sharing. Any other such tutorials would be more than welcomed but I appreciate the amount of work it must have taken to produce.
Very informative. I've always wondered what they ment by DCOs. I used to think it was a marketing buzzword to make a digital oscillator appear analog, but I stand corrected! Thanks!
Excellent video. This would be equally important to an electronics theory class or to a history class. Thank you for spending the time. Your explanations are very clear and understandable.
I know I'm just echoing others here, but this was a great explanation. It's really nice learning how things like this work; this is pretty different from how I'd imagined it!
Excellent video!! I'm studying the Poly61 and this really helped me understand how both DCO1 & DCO2 work in the poly. I wish you could do the poly as well, the schematics are really not that bad. ;)
This is pretty much the best video ever made. Now I know why my Juno-6 has “life” even without the chorus. I’m curious about the DCO design of my JX-3p and if the master clock is crystal or not.
Would be nice if in the future you could show people the schematics to a Roland SH-101 say maybe surprised to see what the inner workings in circuit are ;) ... Thank you for a great and informative video
Hi GOLT! Great Job. Did you actually verify the shape of the pulse from the 8253 to the SAW-Wave-Shaper of the JUNO 06 with an oscilloscope? As far as I understand, the 8253 outputs a short negative pulse, that is used to to shortly switch on the PNP TR5, which in turn discharges the capacitor to 0V. Then the positive voltage from the S&H 1/2 IC35 sends a current into R36, which in turn makes makes TP3 go negative. So the sawtooth goes down from 0V to -12V. What do you think?
Brilliant , this vid should be shared in masses so those 106 lovers finally start to sell their .... DIGITAL Synths :) and these ridiculous prices will drop . haha
The Juno 106 still has a wonderful analogue filter, I consider it a hybrid synth actually. But you are right the prices are insane. But the new Behringer might fix that. It should be a better synth than the 106.
What is the name of the synthesizer type, if it uses digital sound generation and analog filters? Alpha Juno is hybrid ? Or analog? Can DCO generate a wave digitally (like Alpha Juno)? Or this type of oscillator will no longer be considered DCO? (despite the fact that the manufacturer called it the DCO in manual) ...but it is not indicated anywhere that the DCO must necessarily be analog ... digital in both cases only control the pitch
I finally had the time to watch this video. Thanks very much for this great content! What about some future tech videos? I'm thinking of VCFs, VCAs,... :-).
Fabulous video. I'm not an electronics guy, but I'm planning on buying the DeepMind and already have a Sub 37 so I have been very interested in getting the lowdown on just what DCOs do.
Very helpful, thanks! Though that can of worms is interesting: what do you take as the definition of analogue vs digital? I'd always thought that "digital" meant that at some stage the waveform is represented as a series of numbers, which then get converted to voltage via a DAC. In this case, even though there's a meaningful difference between the master clocks, the voltages that get shaped and then filtered are always continuous voltages that come from (dis)charging capacitors. So, even though the quirks of the clock division used to reset the ramp can create artefacts due to not being able to represent precise frequencies, there will never be aliasing or low-bit-depth artefacts. Does that make sense?
That makes sense. All these synths I consider to be analogue. But the oscillator for the most part is not, it's a calculated frequency. But at the same time you don't get aliasing which is a digital trait. Really they are all hybrids to varying degrees, and in hindsight I should have said that in the video. Although I think the best part about analogue is gone, which is the instability.
+Eman Mann I agree. It now depends on processing power and the knowledge and time the software developer has. I have owned many Virtual Analogue synths and the part I find most annoying is the pitch variation on the oscillators is never truely random like an old analogue. They just apply standard digital modulation waveforms, which are all highly reproducible, and even the same modulation source is applied to all oscillators! To do it properly you would need to apply different random modulations to each oscillator individually. That takes too much processing power in most cases. They need something quick to write on the marketing advertisement, not a "real" emulation.
is there any information/literature around just how the current changes for an oscillator? or like the nature of "random" drift in analog circuits in general?
+0e0 Most of the drift attributed to temperature is caused by a circuit called the exponential convertor. The control for a VCO is a 1V/Oct linear scale in most synths (except in Hz/Volt synths) but oscillators need a non-linear "double the current = double the frequency" scale. So there needs to be a circuit that converts the linear Voltage to an exponential current. It's this circuit that causes a lot to the drift. Look up "exponential convertor in VCO". It's more than that though. Analogue VCOs have micro variations that can be seen on a good tuner. Those variations are not present in DCO's. I've had some requests to do a vid on VCOs, so I might do one later.
Thank you! Awesome video! I've got the most of my questions answered! Just one thing I don't understand in details for Jupiter8 VCO - how the expo converter part works.
Awesome video! I'm currently designing a hybrid oscillator that can run in different modes, as an experiment, so this is a relevant refresher. I kept thinking about your statement about the upward versus downward saw/ramp, where you say downward usually contains more low end. Can you elaborate? Why is that? The reset pulse length affects low end, but in both directions equally. So is there another factor at play?
These are great videos and explanations for someone like me who loves science but gets glazed over with electronics. I have a question. What would you call the oscillators on a Sequential Prophet-6 and OB-6 and how are they regulated?
To bad we couldn't go over the Korgs, but otherwise a really great video. I was always under the impression that the Poly 800 had digitally generated waveforms and oscillators (not digitally controlled VCO's) but turns out it's more analog than the Juno 106! Technically anyway
Definitely the AX80. The sound of the AX80 filter is much better than the little korgs in my opinion and it has more options. It's also more highly sort after.
I've owned both and have used both for a long time and the AX has a beautiful apperance with its fluorecent displays and have a unique sound imo. Om the other hand, I've always adored the Poly 61 for its ability to take on lead sounds and pads so they both have merits for different reasons to me. Great tutorial btw! :)
I'm blown away from this information! I'm no engineer, but this is so interesting! I would love to hear your explanation of how VCO's work too. Specifically poly synth VCO's like those in my Korg Minilogue. I'm confused how they hold tune without computers...
Great video! I have a AX80 that has a very weak saw wave around that opamp coming on OSC2 voice 4 .Since the square wave is derived from the saw its also really weak. I'm trying to figure out what's causing the issue at really low frequencies. Mid to high you can hardly tell its out of whack. Could be the input cv, the transistor, or a weak cap.
Great video, I wish I could like it twice. I'm working on my Poly-61 and I have almost finished redrawing the digital board schematics. Would you be interested in recording a follow-up video about Poly-61 if the schematics were readable? I'd like to create a replacement digital board and need to understand a few more things (e.g. what is the anti log circuit doing) to keep 61's character while removing its limitations.
Hey! Thanks for the video! Do you offer any support on juno 106 diagnostics? I have changed all 6 "voice" chips and 3 wave generator chips for new ones, but still have completely no sound on voice #5. Voice #6 outputs only the sub and the noise + I can hear that it kinda tries to make the sound (super fast attck and super fast release) for the saw and pulse. Also did the filter test. On voice #5 I cannot hear that self oscillation. Other voices are good. Do you know what could have fone wrong here?
Thanks for your video on this topic, I was eagerly awaiting your description of the Poly 61 DCO as I have no less than 3 synths in my workshop with faulty voice boards awaiting repair. The VCF's contain a lot of discrete components making for a complex circuit compared to the Poly Six. As you mentioned the schematics are generally very poor quality making life harder...anyway I'll stop complaining now. My question is whether you could recommend any good texts that cover the circuit design principles? CMOS cookbook?
Thanks for this fantastic video, I have learnt so much from this. I have a Juno 60 which died recently (CPU failure, now repaired) and to help with the repair, I borrowed a friend's Juno 6. I have always felt that the DCOs on my 60 have a very stiff, almost unforgiving quality about them. I was surprised to find that my friend's Juno 6 had quite a different quality; the DCOs were, in fact, slightly out of tune with each other. When playing octaves, I could clearly hear that there was some slight detune happening, which made the Juno 6 sweeter and more 'relaxed' sounding compared to my hard and precise Juno 60. I have always wondered how this could be the case when they are supposedly identical voice boards with only the one master clock present, as you have explained in your video. Both of the 8253 counter chips are receiving this same 1.90MHz clock. Could it be tolerance differences between the two 8253 chips, causing slight differences in tuning between voives 1,2,3 and 4,5,6? Or could this be something happening more at the saw generation level, or even the key CV value? Would be keen to hear your thoughts!
You are right! ( I have a strong opinion about this ) - Sorry it has taken me so long to reply. I actually tried to reply as soon as you posted the message but it failed to send, and then I forgot for a while! Firstly I completely agree with you. The Juno 6 is my favourite of the Juno's and there is definitely a reason why it sounds more "relaxed". I don't think it's the counter chips. The counter chips are the same as the 60 and should divide the same. I think it's the CVs and it is a huge difference. In the 60 all the CVs are quantised into 256 steps by the CPU before going to the analogue circuits. In the Juno 6 they DO NOT travel through a CPU, they are all immediately applied in all their naturally variable analogue glory. This greatly effects the synth. You can notice in so many ways, even in the filter it's obvious. Turn the resonance up to full in the 60, then slowly move the filter cutoff, you can hear the stepping in the filter due to the CPU. Do the same on the 6 and the stepping is gone. All the control voltage are effected, so even if the DCO tuning is similar, by the time the voice makes it's way to the output, it's clear the 6 is more organic.
+Falk Hoffmann (falconi) The falling saw sounding fuller is just my opinion based on Eurorack oscillators that allow you to shape between the two (like RS95E). And combining that with my general opinion that the synths I the think sound fuller in the low end, tend to have falling saws. It's only my opinion and I could be wrong. It could be that the actual oscillator design has more impact than the saw direction. I don't have a good spectrum analyser so I can't verify any of that.
Thank you for this very informative reply, a total revelation! I had heard many times about the Juno 6 being a bit 'warmer' from various people and I always sort of dismissed it until I actually compared my Juno 60 to my friend's 6. Those CVs getting applied to the voices directly on the Juno 6 makes a lot of sense and solves the mystery. Thank you again!
Falk Hoffmann I know this is months after your reply but I wanted to chime in. I was recently doing some expirements on this very subject and came to the same conclusion. That ramp vs saw have a different bit of low end to them. What was odd is using a spectrograph I saw (pardon the pun) no difference in frequency content. I've heard rumors that it's a psychological effect dealing with the way we humans perceive sound though. I'm still looking into this. Ethan Winer seems to think he has it figured out here. ethanwiner.com/ear_imd.htm
Interesting article, thank you. I'd suppose, that a possible fuller sound (of ramp down-oscillators) is NOT dependent on absolute polarity. Imho, it should be a result of the circuit design and its output signal - and it must be visible in the spectrum, of course...but maybe, I'm wrong. Did you digitally record the signals from your experiments? Did you later (phase-)invert the signals in a wave editor software and compare them to the originals? Did you invert the polarity of your speakers, if possible, and repeat the test? Did you carry out experiments with high-pass filtering at very deep frequencies (around 10 to 50Hz)? How does your experimental listening and recording setup look like? Analog synth or self-built oscillator on a pure analog signal chain? Or Software (e.g. Reaktor etc.)? May your audio-interface play back/record very deep frequencies, so is it "almost DC-coupled"? Just ideas from my side. Both analog and digital oscillators may cause a lot of rumble...
This is a great video. Thank you! What does it mean for a JX-3P. As I understand it the envelopes on the JX are generated by the CPU, but I don't know if the original master clock is crystal or analog since its design is older than the 106.
Great explanation!! One of the reasons I love the clavia modulars is you can replicate exactly what is described here... master clocked osc, with as many slave osc's your patch can fit into memory! :)
Great analysis, thanks for the video, I am trying to reproduce the juno core on a breaboard. What should be the value of C7 (the capacitor "responsable" for the saw) ? The schematics looks like 0.001 G, can't read the unit or I should say can't make sense out of it.
Hello, I plan to bluild a DCO for musical application from 16hz to 8000hz with a sawtooth output signal. My design of the DCO works the following way by managing two things from the Microprocessor (inspired from Juno 106): •The sawtooth positive slope by giving the correct voltage to a RC network to have the adequate voltage at the end of the note period. •The discharge event when the microprocessor immédiately shortcut the RC network to restart the next sawtooth period from 0v at the right time. How to manage the discharge Event with this ATTiny85? For the discharge event I have to count from an external oscillator a certain number of clock cycle with the microprocessor. I wanted to use the common and cheap ATTiny85. My problem is that it contain two 8bits counters but none 16bits. For that i Identified two technics: •Cascading the 8bits counter using the overflow flag, but it seems that from my reading that i would introduce counting errors as it takes time for the µP to detect the flag and store the value. •Another solution would be to pre-scale the second counter (fclockin/1024) and count the clock divided by 1024 but it seems that prescaling is not allowed from external oscillator. But devil is in detail, i have also to manage and external clock prescaler to divide the 8MHz clock by 2,4,8,16 to have a better frequency résolution at low frequency notes. But, that should not be a big issue. What would you suggest me? Any other way to do it? Thank you for your help!
It's a pity you didn't do the Korgs because I've heard that the Korg Poly 800 doesn't use a capacitor to generate the saw wave, but rather generates a simple stepped saw by mixing multiple octaves of square wave in the correct ratios. I have no idea if it's true, I watched this video hoping to learn if that was the case.
Awesome! It seems to me that all that really matters when it comes to the resulting waveform output from any osc. is if it allows further processing without aliasing. Correct?
My Cello teacher once said to me, "if it sounds good, it is good". So really it doesn't matter what oscillators a synth has. It's still fun to learn more about them though. I tried to not judge them, but sometimes you find yourself judging them anyway!
Hi Great video. Am tying to fix an early poly 61 and would like to understand how it works. any Chance you can talk us through the diagrams as I`m lost with it? Thanks in advance. :-)
Great video! I actually thought the 106 had a DC-controllable oscillator, since it has a tuning trimmer on the back. How did they manage to do that, when the oscillator is crystal clocked?
My understanding is it's done the same way the bender and LFO effect the frequency, the CPU just calculates a new divider ratio. There is no way to change the crystal clock, it's impossible, so it must be done digitally.
In the JX-3P, a CV from Tune pot (which I could neither find in the block diagram nor the schematic) and Bender (it's there!) are A/D-converted before being interpreted by the CPU (obviously running at 12MHz). With these data, CPU seems to digitally finetune a master oscillator circuit with a center frequency of approximately 6MHz (with three dividers, like in the 106), later clocking all 12DCOs.
Amazing as always! I heard a myth that Poly 61 actually misleading with "DCO"s and got 1 discrete analog oscillator and 2nd pure digital (they are really sounds VERY different) - but it could be only a myth though :) Is Kawai SX210 uses the same approach as 240?
Yes the Poly-61 has purely digital oscillators as well. The DCO-2 saw is a stepped digital waveform with quite bad resolution. But that doesn't mean it sounds bad. People can make up there own minds. DCO-1 is the more traditional one. I had this discussion with a friend recently, technically any digital waveform is a DCO, because they are digitally controlled. It can be hard to tell when the "marketing" takes over from reality.
Yes. Poly-61 actually sounds very good on basses because of the 12db low pass, not everything depends on the oscillators :) Just checked SX210 service manual - and yes its exactly the same 4mhz crystal (they call it "MASTER OSCIRATER" - not sure about the rest though :) Will we see a Kawai repair soon? Thanks for your amazing work!
would you know anything about repairing a korg poly 61 that has been damaged from a leaked battery? and if so would you be willing to fix it for some $$$?
struggling with this, not because of the video.. I'm just dense sometimes. so in other words.... my simple understanding between the 2 was that a digital clock something or other with the oscillator.. DCO. vs with VCO it's voltage passing through at a specific..rate?hz?
Great video! I just have one small disagreement; I don't think the difference between the clocks used in the Juno-6/60 and the 106 is as big as you make it out to be. Any error in the VCOs is divided by ~1000, which would make even severe drifting in the VCO virtually inaudible. I don't think you can attribute much of the sonic differences between the 6/60 and 106 to this.
Johan Fjeldtvedt I thought about that when I did the video, but what changed my mind was that the clocks were analogue SO they could be voltage controlled (by as much as an octave). The clock input CVs were able to shift the pitch by large amounts and those input CVs weren't precise or that well controlled. Granted they are not as unstable as true VCOs but the fact that there are quite dramatic input CVs changed my mind. They will not have the same stability of DCOs clocked by crystals because of the input CVs I think. The crystal based DCOs had the bender/LFO CVs quantised by the CPU first, so there is no way those small shifts will be present. That's my theory anyway. :-)
Any degree of variation in the clock period will also be in the note periods, as the note periods are created in the circuit by adding together multiple clock periods.
Hi, GOLT! If you could explain this... Do DCOs triggers the wave on every note or DCO can work as a "free wave" oscillator? As I know usually VCO works continuously and just shaped by VCA. This questions seems silly for me but I really not sure if it "plays the wave" in the same way as VCO.
You might record your RS95E in your DAW while fading from UP to DOWN and later analyze the spectrum in Audition/WaveLab/one or the other PlugIn. Funfunfun!
O.K.. I just had a magical 40 minutes! Although I have been working with synths for decades and also started to repair and convert them years ago I still never had a moment when everything totally fell into place. Now that moment has come. Hurray, I at last understand how oscillators work on a component level! Suddenly all earlier, rather random insights fall into a linear tale. So from now on it will therefore be much easier to juggle all that information around. It's comparable to when somebody first made all the bits about music theory come together for me. I can still remember that magical moment as I will remember this one. So thank you very, very, very much!
You're welcome Marc. Glad you enjoyed it.
Yeah! Sorry if that sounded pompous but it is the actual truth. That is the nice side of the internet. Someone from the other side of the world (I'm Dutch!) can really make a difference because he simply has the inclination to share his knowledge without a preconceived plan for personal (economic) gain. So I'm affraid I'll have to repeat it again: THANKS!
I add my voice to Marc. With decades of playing and fixing synths, I thought I was understanding DCO, but your video confirmed I was not. You lighted up all the black spots I had. Many thanks for that. However, the dark side of this video is .... we want more! ;-)
Great video, really enjoyed watching that. You kinda guessed right - the DeepMind12 has 100MHz master oscillator and is divided down to reach low pitch using a combination of 32 bit and 16 bit timers, plus we're using higher res DACs so don't need the range switch for the DCO CV.
Rob Belcham is that a comment from one of the Deepmind's developers? :)
Rob Belcham ha fancy to see your comment here. This was a great video and nice to see you chiming in so I don't have to twist your ear about it in chat lol! Such a great video
I'm not an electronics guy, and still I understood enough of your explanation to really learn loads. Thanks!
I'm such a fan of your channel! One of the most informative, comprehensive synth repair channels anywhere. Thanks for taking time to share and educate us synth lovers. I have a friend's Juno 106 in my studio that I'm trying to repair and the poor thing is pretty bad. Wish me luck, your videos really help!
Wow! Thank you for this concise and thorough explanation of the master clocks! I've always loved the sound of the Poly61, and felt that the 106 sounded a little too "tame". Now I fully understand why!
Exactly the kind of synth geekery we need! Already the second time I watched this. Thanks for the great content!
Brilliant - love it. First time I have seen this actually explained.
Very interesting, thanks for making this. Watching it for a second time again
This video was wonderful. You did such an excellent job explaining all of this. If you're not a teacher, I think you would have been a fantastic one. Thank you so much for making this video.
this was an amazing analysis, I wish we could see more of these!
Thanks for a very interesting and informative lesson! I really enjoyed this video, and - oops - just realized that I watched it all with the sound going through the Elektron Analog Heat sound processor set to the "Saturation" tape-like gentle distortion circuit. I love the oldschool tape recorder instruction movie feeling!
Cheers! ;-)
Fantastic! Thank you for expanding my understanding!
What an excellent video! Helped me not only to understand DCOs but also VCOs! BTW: Are you a teacher? You have very good skills of explaining stuff!
wonderful tour of these schematics, thank you for taking the time to make this. a few points stand out that are worth noting.
1: thinking about a pnp as "normally closed" can get you into trouble. BJTs are all normally open!
2: in both cases of rising and falling saws, the rate is set by the current into the integrator circuit (op amp with cap out to negative input). the resistor in series with the transistor sets the discharge rate and limits the current through the transistor.
3: keeping the sawtooth level constant at different frequencies is the job of the microcontroller, and inherently tied to the ramp rate. this is done with the dacs.
3: rising and falling sawtooths sound identical, this is a fundamental truth of waveforms and harmonics.
4: Poly 800 doesn't have DCOs like this, but instead builds all of its waves by summing pulses!
Ha that was great. I never considered the impact of using a voltage controller osc as a clock for a DCO. Very cool and thanks for sharing this knowledge. Wow, and then you can modulate this clock source. Eye opening!
Excellent and timeless video! I was sent here by JX3D on the ranzee discord. Even with a good level of understanding electronics, I think you have just doubled my understanding of synth circuits in the space of one video. Thank you for 'knowledge bomb' (even if I am 6 years late).
You are so talented. Thanx.
You are the best for us synth geeks who are also our own techs; those of us who really want the details and some great ideas for mods. Thank you for showing us so much. Frank S
Truly excellent video ! Extremely interesting !
What an absolutely FANTASTIC video. You are _amazing_ at explaining electronics. Utterly unique. Thank you!
Really interesting thanks! How can a falling saw sound different to a rising saw? Mr Fourier might disagree.
Amazing. Simply amazing, sir! I do look forward to your video uploads, as they are all so good, and not only interesting, but very educational. Like so many others following, I now can appreciate oscillator circuits better - I can't thank you enough!
Thank you Andy Jackson that's a really nice thing to say. I certainly have some more videos planned.
That's good to hear! I'll look forward to those! Now, back to wrestling this broken OB-Xa (aren't they all?!!) into submission... Ha!
Kaleb, I can't believe I watched this whole thing! This was So interesting and well presented. Thx for spending the hours to put this together and sharing.
GOLT! I wish I could spend a month with you. Great teaching. Thanks!
Superb educational video! Thank you so much for sharing. Any other such tutorials would be more than welcomed but I appreciate the amount of work it must have taken to produce.
Very informative. I've always wondered what they ment by DCOs. I used to think it was a marketing buzzword to make a digital oscillator appear analog, but I stand corrected! Thanks!
Excellent video. This would be equally important to an electronics theory class or to a history class. Thank you for spending the time. Your explanations are very clear and understandable.
This is so fantastic. I have a hugely limited understanding of electronics but this was so clear; you're a great educator.
I know I'm just echoing others here, but this was a great explanation. It's really nice learning how things like this work; this is pretty different from how I'd imagined it!
Really really enjoyed this! Easy to understand and the visuals made the difference!! Thank you for taking time to do this!
Excellent video!! I'm studying the Poly61 and this really helped me understand how both DCO1 & DCO2 work in the poly. I wish you could do the poly as well, the schematics are really not that bad. ;)
Absolutly excellent video explanation. Serious education that has cleared up alot of myths for me .
This is pretty much the best video ever made. Now I know why my Juno-6 has “life” even without the chorus. I’m curious about the DCO design of my JX-3p and if the master clock is crystal or not.
Thank you! You've answered many questions I've wondered for many years.
This video is super informative! Thank you very much for sharing your knowledge.
Would be nice if in the future you could show people the schematics to a Roland SH-101 say maybe surprised to see what the inner workings in circuit are ;) ... Thank you for a great and informative video
SH-101 is a pretty simple synth. You should have no trouble reading the schematics
Excellent video. Very informative. Thank you for your time and effort.
WoW! Thanks for making this. Looking forward to all the rest vids. Keep it up mate!
Hi GOLT! Great Job. Did you actually verify the shape of the pulse from the 8253 to the SAW-Wave-Shaper of the JUNO 06 with an oscilloscope? As far as I understand, the 8253 outputs a short negative pulse, that is used to to shortly switch on the PNP TR5, which in turn discharges the capacitor to 0V. Then the positive voltage from the S&H 1/2 IC35 sends a current into R36, which in turn makes makes TP3 go negative. So the sawtooth goes down from 0V to -12V. What do you think?
Brilliant , this vid should be shared in masses so those 106 lovers finally start to sell their .... DIGITAL Synths :) and these ridiculous prices will drop . haha
The Juno 106 still has a wonderful analogue filter, I consider it a hybrid synth actually. But you are right the prices are insane. But the new Behringer might fix that. It should be a better synth than the 106.
So many questions answered! Excellent video :)
one of the best things ive ever seen on youtube, thanks
What is the name of the synthesizer type, if it uses digital sound generation and analog filters? Alpha Juno is hybrid ? Or analog? Can DCO generate a wave digitally (like Alpha Juno)? Or this type of oscillator will no longer be considered DCO? (despite the fact that the manufacturer called it the DCO in manual) ...but it is not indicated anywhere that the DCO must necessarily be analog ... digital in both cases only control the pitch
I finally had the time to watch this video. Thanks very much for this great content! What about some future tech videos? I'm thinking of VCFs, VCAs,... :-).
Fabulous explanation, with excellent detail. Thank you!
You are a master. Thank you so much. Was a pleasure to watch.
Fabulous video. I'm not an electronics guy, but I'm planning on buying the DeepMind and already have a Sub 37 so I have been very interested in getting the lowdown on just what DCOs do.
Thank you for letting us pick your brain. Very very helpful
You need to setup a Patreon for these videos. These explanations are extremely valuable to people learning circuits. Please?
Amazing explanation!. Wish you’d taught at my university!
Excellent video. Thank you very much, learned a lot!
Fascinating and nicely presented. Many thanks.
Very helpful, thanks! Though that can of worms is interesting: what do you take as the definition of analogue vs digital? I'd always thought that "digital" meant that at some stage the waveform is represented as a series of numbers, which then get converted to voltage via a DAC. In this case, even though there's a meaningful difference between the master clocks, the voltages that get shaped and then filtered are always continuous voltages that come from (dis)charging capacitors. So, even though the quirks of the clock division used to reset the ramp can create artefacts due to not being able to represent precise frequencies, there will never be aliasing or low-bit-depth artefacts. Does that make sense?
That makes sense. All these synths I consider to be analogue. But the oscillator for the most part is not, it's a calculated frequency. But at the same time you don't get aliasing which is a digital trait. Really they are all hybrids to varying degrees, and in hindsight I should have said that in the video. Although I think the best part about analogue is gone, which is the instability.
+Eman Mann I agree. It now depends on processing power and the knowledge and time the software developer has. I have owned many Virtual Analogue synths and the part I find most annoying is the pitch variation on the oscillators is never truely random like an old analogue. They just apply standard digital modulation waveforms, which are all highly reproducible, and even the same modulation source is applied to all oscillators! To do it properly you would need to apply different random modulations to each oscillator individually. That takes too much processing power in most cases. They need something quick to write on the marketing advertisement, not a "real" emulation.
is there any information/literature around just how the current changes for an oscillator? or like the nature of "random" drift in analog circuits in general?
+0e0 Most of the drift attributed to temperature is caused by a circuit called the exponential convertor. The control for a VCO is a 1V/Oct linear scale in most synths (except in Hz/Volt synths) but oscillators need a non-linear "double the current = double the frequency" scale. So there needs to be a circuit that converts the linear Voltage to an exponential current. It's this circuit that causes a lot to the drift. Look up "exponential convertor in VCO". It's more than that though. Analogue VCOs have micro variations that can be seen on a good tuner. Those variations are not present in DCO's. I've had some requests to do a vid on VCOs, so I might do one later.
Thank you! Awesome video! I've got the most of my questions answered! Just one thing I don't understand in details for Jupiter8 VCO - how the expo converter part works.
Awesome video! I'm currently designing a hybrid oscillator that can run in different modes, as an experiment, so this is a relevant refresher. I kept thinking about your statement about the upward versus downward saw/ramp, where you say downward usually contains more low end. Can you elaborate? Why is that? The reset pulse length affects low end, but in both directions equally. So is there another factor at play?
These are great videos and explanations for someone like me who loves science but gets glazed over with electronics. I have a question. What would you call the oscillators on a Sequential Prophet-6 and OB-6 and how are they regulated?
To bad we couldn't go over the Korgs, but otherwise a really great video. I was always under the impression that the Poly 800 had digitally generated waveforms and oscillators (not digitally controlled VCO's) but turns out it's more analog than the Juno 106! Technically anyway
This is just a special case of analog DCOs. But what about digital DCO? (Casio CZ-101, Kawai K1, Kawai K4 etc)
thanks for this golt! perfect timing, as im currently trying to choose between an AX80 or poly-61.
Go Ax , P61 is a VERY limited synth.
Definitely the AX80. The sound of the AX80 filter is much better than the little korgs in my opinion and it has more options. It's also more highly sort after.
I've owned both and have used both for a long time and the AX has a beautiful apperance with its fluorecent displays and have a unique sound imo. Om the other hand, I've always adored the Poly 61 for its ability to take on lead sounds and pads so they both have merits for different reasons to me. Great tutorial btw! :)
I'm blown away from this information! I'm no engineer, but this is so interesting! I would love to hear your explanation of how VCO's work too. Specifically poly synth VCO's like those in my Korg Minilogue. I'm confused how they hold tune without computers...
Jupiter 8 holds tune with Z80 processor... so basically it's "a computer" :) I guess Minilogue can work in the same way.
Great video! I have a AX80 that has a very weak saw wave around that opamp coming on OSC2 voice 4 .Since the square wave is derived from the saw its also really weak. I'm trying to figure out what's causing the issue at really low frequencies. Mid to high you can hardly tell its out of whack. Could be the input cv, the transistor, or a weak cap.
Fascinating. Thanks for making this.
This is nerdy, detailed and really cool. :) Do you know how the DSI/Sequential Prophet Rev2 masterclock is generated?
Great video, I wish I could like it twice. I'm working on my Poly-61 and I have almost finished redrawing the digital board schematics. Would you be interested in recording a follow-up video about Poly-61 if the schematics were readable? I'd like to create a replacement digital board and need to understand a few more things (e.g. what is the anti log circuit doing) to keep 61's character while removing its limitations.
Hey! Thanks for the video! Do you offer any support on juno 106 diagnostics? I have changed all 6 "voice" chips and 3 wave generator chips for new ones, but still have completely no sound on voice #5. Voice #6 outputs only the sub and the noise + I can hear that it kinda tries to make the sound (super fast attck and super fast release) for the saw and pulse. Also did the filter test. On voice #5 I cannot hear that self oscillation. Other voices are good. Do you know what could have fone wrong here?
Great video. It would seem by BITone is a Digital Oscillator synth as it seems to lack the capacitor etc part of the analogue circuit.
This is fantastic, great teacher !
Great video and very good explanation , thanks.
Great stuff man! Exellent. It is a pity that there seems to be no high-quality schematics for the Korg Poly-61 on the internet. Or is there?
If you made a similar video on the IR309 filter that would be amazing!
Thanks for your video on this topic, I was eagerly awaiting your description of the Poly 61 DCO as I have no less than 3 synths in my workshop with faulty voice boards awaiting repair. The VCF's contain a lot of discrete components making for a complex circuit compared to the Poly Six. As you mentioned the schematics are generally very poor quality making life harder...anyway I'll stop complaining now.
My question is whether you could recommend any good texts that cover the circuit design principles?
CMOS cookbook?
Thanks for this fantastic video, I have learnt so much from this. I have a Juno 60 which died recently (CPU failure, now repaired) and to help with the repair, I borrowed a friend's Juno 6. I have always felt that the DCOs on my 60 have a very stiff, almost unforgiving quality about them. I was surprised to find that my friend's Juno 6 had quite a different quality; the DCOs were, in fact, slightly out of tune with each other. When playing octaves, I could clearly hear that there was some slight detune happening, which made the Juno 6 sweeter and more 'relaxed' sounding compared to my hard and precise Juno 60. I have always wondered how this could be the case when they are supposedly identical voice boards with only the one master clock present, as you have explained in your video. Both of the 8253 counter chips are receiving this same 1.90MHz clock. Could it be tolerance differences between the two 8253 chips, causing slight differences in tuning between voives 1,2,3 and 4,5,6? Or could this be something happening more at the saw generation level, or even the key CV value? Would be keen to hear your thoughts!
You are right! ( I have a strong opinion about this ) - Sorry it has taken me so long to reply. I actually tried to reply as soon as you posted the message but it failed to send, and then I forgot for a while! Firstly I completely agree with you. The Juno 6 is my favourite of the Juno's and there is definitely a reason why it sounds more "relaxed". I don't think it's the counter chips. The counter chips are the same as the 60 and should divide the same. I think it's the CVs and it is a huge difference. In the 60 all the CVs are quantised into 256 steps by the CPU before going to the analogue circuits. In the Juno 6 they DO NOT travel through a CPU, they are all immediately applied in all their naturally variable analogue glory. This greatly effects the synth. You can notice in so many ways, even in the filter it's obvious. Turn the resonance up to full in the 60, then slowly move the filter cutoff, you can hear the stepping in the filter due to the CPU. Do the same on the 6 and the stepping is gone. All the control voltage are effected, so even if the DCO tuning is similar, by the time the voice makes it's way to the output, it's clear the 6 is more organic.
+Falk Hoffmann (falconi) The falling saw sounding fuller is just my opinion based on Eurorack oscillators that allow you to shape between the two (like RS95E). And combining that with my general opinion that the synths I the think sound fuller in the low end, tend to have falling saws. It's only my opinion and I could be wrong. It could be that the actual oscillator design has more impact than the saw direction. I don't have a good spectrum analyser so I can't verify any of that.
Thank you for this very informative reply, a total revelation! I had heard many times about the Juno 6 being a bit 'warmer' from various people and I always sort of dismissed it until I actually compared my Juno 60 to my friend's 6. Those CVs getting applied to the voices directly on the Juno 6 makes a lot of sense and solves the mystery. Thank you again!
Falk Hoffmann I know this is months after your reply but I wanted to chime in. I was recently doing some expirements on this very subject and came to the same conclusion. That ramp vs saw have a different bit of low end to them. What was odd is using a spectrograph I saw (pardon the pun) no difference in frequency content.
I've heard rumors that it's a psychological effect dealing with the way we humans perceive sound though. I'm still looking into this. Ethan Winer seems to think he has it figured out here. ethanwiner.com/ear_imd.htm
Interesting article, thank you.
I'd suppose, that a possible fuller sound (of ramp down-oscillators) is NOT dependent on absolute polarity. Imho, it should be a result of the circuit design and its output signal - and it must be visible in the spectrum, of course...but maybe, I'm wrong.
Did you digitally record the signals from your experiments?
Did you later (phase-)invert the signals in a wave editor software and compare them to the originals? Did you invert the polarity of your speakers, if possible, and repeat the test?
Did you carry out experiments with high-pass filtering at very deep frequencies (around 10 to 50Hz)?
How does your experimental listening and recording setup look like? Analog synth or self-built oscillator on a pure analog signal chain?
Or Software (e.g. Reaktor etc.)?
May your audio-interface play back/record very deep frequencies, so is it "almost DC-coupled"?
Just ideas from my side.
Both analog and digital oscillators may cause a lot of rumble...
This is a great video. Thank you! What does it mean for a JX-3P. As I understand it the envelopes on the JX are generated by the CPU, but I don't know if the original master clock is crystal or analog since its design is older than the 106.
Great explanation!! One of the reasons I love the clavia modulars is you can replicate exactly what is described here... master clocked osc, with as many slave osc's your patch can fit into memory! :)
Great analysis, thanks for the video, I am trying to reproduce the juno core on a breaboard. What should be the value of C7 (the capacitor "responsable" for the saw) ? The schematics looks like 0.001 G, can't read the unit or I should say can't make sense out of it.
Hello,
I plan to bluild a DCO for musical application from 16hz to 8000hz with a sawtooth output signal.
My design of the DCO works the following way by managing two things from the Microprocessor (inspired from Juno 106):
•The sawtooth positive slope by giving the correct voltage to a RC network to have the adequate voltage at the end of the note period.
•The discharge event when the microprocessor immédiately shortcut the RC network to restart the next sawtooth period from 0v at the right time.
How to manage the discharge Event with this ATTiny85?
For the discharge event I have to count from an external oscillator a certain number of clock cycle with the microprocessor. I wanted to use the common and cheap ATTiny85. My problem is that it contain two 8bits counters but none 16bits.
For that i Identified two technics:
•Cascading the 8bits counter using the overflow flag, but it seems that from my reading that i would introduce counting errors as it takes time for the µP to detect the flag and store the value.
•Another solution would be to pre-scale the second counter (fclockin/1024) and count the clock divided by 1024 but it seems that prescaling is not allowed from external oscillator.
But devil is in detail, i have also to manage and external clock prescaler to divide the 8MHz clock by 2,4,8,16 to have a better frequency résolution at low frequency notes. But, that should not be a big issue.
What would you suggest me? Any other way to do it?
Thank you for your help!
Thanks for taking time to explain this once for all. Does Roland Jx series have analog or digital master clock? keep up good work!
It's a pity you didn't do the Korgs because I've heard that the Korg Poly 800 doesn't use a capacitor to generate the saw wave, but rather generates a simple stepped saw by mixing multiple octaves of square wave in the correct ratios. I have no idea if it's true, I watched this video hoping to learn if that was the case.
Thank you very much , amazing video
Why do some DCO synths still need to be warmed up in order to calibrate?
hi professor could you explain the dcos on the alpha Juno?
Awesome! It seems to me that all that really matters when it comes to the resulting waveform output from any osc. is if it allows further processing without aliasing. Correct?
My Cello teacher once said to me, "if it sounds good, it is good". So really it doesn't matter what oscillators a synth has. It's still fun to learn more about them though. I tried to not judge them, but sometimes you find yourself judging them anyway!
So my roland jx8p which is a dual dco has only rising waveforms?
Very interesting and well explained. Thanks! By the way, is the Jx-3P like the 106 and with a crystal master osc?
Voice 1 and 2 of a JX-3P were controlled by analog inductive masters at 6mhz, you could tune them with screwdriver.
I feel sorry for those poor capacitors being shorted out so often. Component quality is key it seems to me.
Hi Great video. Am tying to fix an early poly 61 and would like to understand how it works. any Chance you can talk us through the diagrams as I`m lost with it? Thanks in advance. :-)
Great video! I actually thought the 106 had a DC-controllable oscillator, since it has a tuning trimmer on the back. How did they manage to do that, when the oscillator is crystal clocked?
My understanding is it's done the same way the bender and LFO effect the frequency, the CPU just calculates a new divider ratio. There is no way to change the crystal clock, it's impossible, so it must be done digitally.
You can change the frequency of a Crystal oscillator .....if you shunt it with a variable capacitor.
In the JX-3P, a CV from Tune pot (which I could neither find in the block diagram nor the schematic) and Bender (it's there!) are A/D-converted before being interpreted by the CPU (obviously running at 12MHz).
With these data, CPU seems to digitally finetune a master oscillator circuit with a center frequency of approximately 6MHz (with three dividers, like in the 106), later clocking all 12DCOs.
Amazing as always! I heard a myth that Poly 61 actually misleading with "DCO"s and got 1 discrete analog oscillator and 2nd pure digital (they are really sounds VERY different) - but it could be only a myth though :) Is Kawai SX210 uses the same approach as 240?
Yes the Poly-61 has purely digital oscillators as well. The DCO-2 saw is a stepped digital waveform with quite bad resolution. But that doesn't mean it sounds bad. People can make up there own minds. DCO-1 is the more traditional one. I had this discussion with a friend recently, technically any digital waveform is a DCO, because they are digitally controlled. It can be hard to tell when the "marketing" takes over from reality.
Yes. Poly-61 actually sounds very good on basses because of the 12db low pass, not everything depends on the oscillators :) Just checked SX210 service manual - and yes its exactly the same 4mhz crystal (they call it "MASTER OSCIRATER" - not sure about the rest though :) Will we see a Kawai repair soon? Thanks for your amazing work!
would you know anything about repairing a korg poly 61 that has been damaged from a leaked battery? and if so would you be willing to fix it for some $$$?
Very interesting video !
Great! Thanks a lot! Very clear...
struggling with this, not because of the video.. I'm just dense sometimes.
so in other words....
my simple understanding between the 2 was that a digital clock something or other with the oscillator.. DCO. vs with VCO it's voltage passing through at a specific..rate?hz?
Fantastic video!
Great teacher ✌🏾
Great video! I just have one small disagreement; I don't think the difference between the clocks used in the Juno-6/60 and the 106 is as big as you make it out to be. Any error in the VCOs is divided by ~1000, which would make even severe drifting in the VCO virtually inaudible. I don't think you can attribute much of the sonic differences between the 6/60 and 106 to this.
Johan Fjeldtvedt I thought about that when I did the video, but what changed my mind was that the clocks were analogue SO they could be voltage controlled (by as much as an octave). The clock input CVs were able to shift the pitch by large amounts and those input CVs weren't precise or that well controlled. Granted they are not as unstable as true VCOs but the fact that there are quite dramatic input CVs changed my mind. They will not have the same stability of DCOs clocked by crystals because of the input CVs I think. The crystal based DCOs had the bender/LFO CVs quantised by the CPU first, so there is no way those small shifts will be present. That's my theory anyway. :-)
Any degree of variation in the clock period will also be in the note periods, as the note periods are created in the circuit by adding together multiple clock periods.
Hi, GOLT! If you could explain this... Do DCOs triggers the wave on every note or DCO can work as a "free wave" oscillator? As I know usually VCO works continuously and just shaped by VCA.
This questions seems silly for me but I really not sure if it "plays the wave" in the same way as VCO.
You might record your RS95E in your DAW while fading from UP to DOWN and later analyze the spectrum in Audition/WaveLab/one or the other PlugIn. Funfunfun!
fantastic ! thank you so much
Awesome video!!
Awesome Vid... Thank you.
amazing knowledge man!!!!