the filter definitely has a big impact of the sound. Listen to demos of a moog 904a (and clones) if can on soundcloud/youtube/whatever, its so bad ass I can't believe it.

But according to people in the know the oscs and mixers and amps all contribute too.

Thanks for the info!

Well, I go on about it quite a bit, but man, the KingKORG sounds glorious and does a great job imitating some classics. Not sure I'd sell anything too cool to get one, but it does squash my G.A.S. for a Prophet 6. It's complement of DWGS waveforms also make it a good imitator of the DW8000. Can it replace it perfectly? Probably not.

As for the source of analog mojo... there are threads and threads on that topic. Will it ever be really answered? Not sure. Even Keanu Reeves figured out he was being represented digitally. I think it's best just to find synths you like and disregard how the sound is generated.

The analog oscillator. It is what it is.

So let's list all the instances where alternative waveforms are available on a non-modular analog synth...

- moogs allow you to blend the typical waveforms together with the waveform knob to form "hybrids"

- the alpha junos have a few different types of PWM sawtooth and some alternative pulse waveforms

- IIRC the Akai AX/VX series have a PWM triangle

- minibrute metalizer

- as I understand it, the Poly-800 sawtooth is actually a stair-stepped pulse wave of some sort

What else? Did I miss anything? Why aren't options like these more common?

The best example I can think of is comparing Serum's metalizer waveform to the Minibrute's version... at least by default Serum's metalizer sounds colder and more controlled, while the Minibrute's is warmer, nastier, and more erratic. I'm sure there are some tricks one could use within Serum to make it seem more "analog", but ultimately then you would be just as guilty of trying to force things out of their wheelhouse. Suffice it to say that waveforms run through the Minibrute have a unique, interesting character which is hard to replicate via wavetable synthesizers.

And if the metalizer can be achieved on such a simple, budget-oriented synth as the Minibrute, just how far out of analog's wheelhouse are these kinds of alternative waveshapes, really? I can't help but think that more analog synths could be exploring waveshaping options like this.

Also, all of the above arguments could be made of Serum's PWM sawtooth compared to the Alpha Juno. Surely these waveforms were born partially out of conveniences of the DCO clock which weren't available on VCO's, but it still seems strange that no one else explored more waveforms like this.

Digital wavetable synths are great, but surely there is some potential for alternative waveshapes in DCO and VCO-based analog synths, due to the alternate innate character of those components.

I was under the impression that NCOs are the technical term for them so Modal just called them what ermm... they're called. I imagine my brother who's an engineer would call them that but i could be mistaken about that.

My point is (if i have one) that it isn't an invented cool marketing term,they're simply called that.

I'm an electronics engineer and have been reading stuff me about sampling and osciillators for thirty years, and the first time I can recall coming across the term NCO was in Modal's blurb.

I guess it's possible I could have missed it up until now, but it's not even a particularly descriptive term, they're no more or less "Numerically Controlled" than a DCO or any other Digital Oscillator (of which they are a form).

In fact the distinguishing feature of them, the variable sample rate, doesn't even require numerical control, you could do that from a VCO if you wanted.

I also note that in the manual of the mono wave (basically the same oscillator design), Paul Maddox referred to them as Digital Oscillators.

They also tried to redefine the term aliasing in one of their explanations, so I'm not really a great believer in them as a source of accurate terminology... fortunately that doesn't seem to prevent them from making good sounding synths.."a rose by any other name" after all

ModCan has an oscillator that is refered to as a NCO and if you google it MicroChip and Lattice Semiconductors has NCOs so i thought it was an old established term. Maybe not.
That i agree with completely. I remember reading about it a few years ago and i was like "Numerical control.....what?"

Paul said 008 was 8 Monowaves in box but i believe it's more like his defender voice board, the PCB shure look like that. Monowave uses a phase accumulator who clocks a external counter that counts adresses from a ROM and phase acc's is no real variable sample rate rather "skip bit" principle albeit the counter acts as a "limited jitter filter".008/Defender voice board uses 3 AT Megas each seams to use the phase accumulators to count internal FLASH for waves if so 008 is quite like non BLIT DW8000 or a ESQ1 unless he uses other principles to count and there are several to chose from. Paul used to talk alot about PPG old phase accumulator design which uses trixes to get around the traditional phase accumulator design limitations.Paul was asked similarely questionas from someone else in the Modulus 008 thread but newer replied , perhaps he didnt know how to answer?!

OSCAr uses "time divison principle" to multiplex out the waveforms so suddenly we go from variable sample rate to fixed sample rate system before waves is processed by the filter but if we assume the time division sample rate is awfully high lets say 2Mhz would you notice?

You're right.

What I should have said is something more like "The first time I encountered the use of the term NCO to refer to the specific characteristics Modal are touting as special"

Ironically NCOs in a Lattice chip appear to be rather more like the more common Digital Oscillators that Modal are specifically trying to differentiate themselves from. They use a fixed sample rate and a variable phase step.

Actually it doesn't, not at the last stage.

Although the sample is taken from the wave memory and converted to analogue using time division multiplexing, The last sample and hold is clocked by the output of a 555 phased locked loop multiplying the output rate of a counter, so the rate at which the output of the oscillator is updated is precisely 128 times the output pitch (256 sample table).

My assumption from what Maddox wrote, and looking at the schematics of the Monowave (Though there is a bit I just can't work out), I think there may be something missing off the schematic I'm looking at, the AT Mega is used to generate a sample clock (with a bit of jitter if they're not using a simple counter since the master clock is a fixed oscilltator), which is the used for an external sample position counter which is always incremented by one each time. So what we get is in effect a variable sample rate, albeit with some jitter.

The ESQ1 on the other hand would skip samples at higher playback pitches, which is rather different. I supposed at lower pitches you could say it is similar to the way the Monowave works, but with way higher jitter of the point when the output sample is changed.

Another way of looking at it would be to say that they are both effectively the same as resampling the output of an ideal variable sample rate oscillator, but the Monowave resamples at a considerably higher sample rate, with reduced aliasing as a result.

I'd like to see a schematic of the defender, I can't find any info on it so far.

Yeah, the challenge seems to be getting inside the midrange of the sound and designing those harmonics.

We can do HP/LP filters w poles to shape the lows/highs, and we can do pulse width to shape some mids, but a very limited number of shapes. What about only even harmonics? What about emphasized minor 7/6 thirds and minor 7/4 7ths with all major-scale harmonic removed? There are infinite timbres that could be matched up with tunings...if we could shape the harmonics in very specific ways.

I see some people posting from the modular world; have any of you guys used the MOTM-E350 Morphing Terrarium for this purpose? Useful?

Not in my schematics, i have 7 pages, hand drawn by Crish Hugget himself in these there is no 555, and last S/H stage (IC27) is clocked by And's (IC 37) who is in the PLL loop with two 4046's a 8253 and a Z80 CTC, defined as sync signals. The CTC counters might act as the wave readout and the 8253 are frequency deciding counters also frequency multiplicated and 4046 is the actual osc's that "ping's" the CTC!? Perhaps you have access to some other schematics?

OSCar is somewhat bizarre in design it really runs on the very edge of what the IC's can do in terms of frequency.

Paul has published all the original complete Monowave schematics, they are out there.

I dont see any particular difference between e.g 008 and ESQ1 both phaseaccs, ESQ1 runs at low speed and limited bitwidth 18 or 20 bit's i recall
and using bandwitlimited wave pages/tabels, 008 runs at 380khz according to rumours and uses 32 bits i "assume" and direct readout no external roms
as per Monowave and also no Monowave type jitter filter unless AT megas has something internal he can use. The jitter is a result of the skipped samples and to low sample rate we know that, Paul mentioned in the other thread he had fizzle on the highest frequencies which suggest he actually runs the Modulus 008 as plain phase accumulators, true variable sample rate would as you say yourself would not produce high en jitter and would also be consistent i wave page length's.

Paul has not published any defender schematics, only a picture of assembled PCB.

Roll your own with Zebra2.

It's always worth looking into sync if you have two VCOs. It's all too common when you have sync to do that screaming Jan Hammer thing, but you can create some beautiful sounding waveforms if you just tweak the sync'd VCO's frequency and keep it static. A sync'd sinewave or triangle wave can be an excellent way to get away from the standard four. You can create almost FM like tones without the usual tuning problem you have analogue FM.

I'm also a fan of waveform mixing - not crossfading like the Voyager et al - but simply mixing one waveform with another in different portions. The SH-101 and Junos do this and it gives both those synths so much more than synths that simply switch between the waves.

Finally, it you have two filters, like the CS30, Blofeld, Virus, etc - filtering one waveform and adding it to another filtered or unfiltered wave is brilliant.

Tony

Sorry, it's the 4046 PLL, not a 555, it's been a while since I looked ar the schematics.

I have the same schematics (looking at them now), i also have Chris' annotated source code somewhere (very generously supplied by him when I was working on the impOSCar) and spent a while probing the hardware in various ways.

The 8253 sets the sample rate, or rather 1/32 the sample rate (the 4046 combined with the 74ls393 counter and a flipfliop in a 74lLS74 form a multiplier.

Ok, now it starts to get really clever.

This sample rate is used to clock one counter in the Z80CTC (the other oscillator the other counter), the value from this is used to read the waveform memory. as you guessed, but the clever bit is that Chris (ab)used the Z80CPU DRAM refresh cycle and line to ensure that the readout of the waveform using the CTC counter to provide the address happens when the CPU isn't going to access it. He also uses the lab of the CPU address line during that refresh cycle to select which oscillator is being updated.

The DRAM refresh stuff was a very useful feature of the Z80 put in there to make it easier and cheaper to use that processor with cheaper DRAM, but Chris didn't need it for that because he was using SRAM, so instead he found a way to make use of it that the Zilog designers almost certainly never considered, very neat.

Anyway, I digress, back to the oscillators.

If you look at sheet 4 (well sheets in my case) you see IC31 which in combination with buffers and capacitors acts as a demultiplexer and sample and hold for the output of the DAC to create control voltages, including the two oscillator volumes.

When it comes time to read an oscillator (still part of the time division multiplexing as you rightly surmised) that oscillator volume voltage is used to set the reference voltage of the DAC (so it acts as a multiplier of the oscillator volume and the waveform) and gets clocked into the sample and hold on the output of IC26

So, at this point the voltage changes at the nearest point after the CTC has updated that the TDM circuitry allows... but notice that this isn't the output.

There is one more sample and hold stage, clocked directly from the circuitry that forms the phase locked loop clocking the CTC. It is independent of the TDM logic, so the final output will actually change at precisely the point that the sample update clock ticks.

So, the oscillator is actually variable sample rate, even though TDM was used to get the voltage there ready to be output.

The oddness of the OSCar circuitry isn't down to operating frequency, everything works inside the range it was specified for, but rather because Chris is brilliantly twisted and found very neat and rather unexpected ways to achieve what he wanted in the minimum amount of circuitry. Pretty much every decision he made makes more sense when you consider he was trying to squeeze the most out of the least, for example the number and types of controls.

I know, I've been looking at them

http://www.elby-designs.com/webtek/m...schematics.pdf

I can't say for certain about the 002 (the 008 is the VCO synth, I assume that was just a typo on your part), but look again at the monowave schematics.

Each oscillator has its own DAC, the input of which is fed from a 64K byte RAM.

The top 8 address bits come from an 8 bit latch, this corresponds to the 256 different waveshapes the manual says are availabe.

The bottom 8 bits come from a counter, they have to be incrementing sequentially, no sample jumping can occur (other than the hirez switch forcing the lower two address bits to zero which reduces the wavetable size to 64 steps rather than 256).

The bit that seems to be missing off that schematic is where the clock to the wave position counter (the CD4040 on the oscillator sheets) comes from, it shows it is the same as A0 for the wave memory, but not where it comes from. However if you look just above it you see a CD4024 counter being used for octave division, with the octave outputs going to a switch, but nothing indicating where the output of the switch the goes to... I'm reasoning that this is A0 for the wave memory and the clock for the wave memory pointer.

So, in the monowave at least it would appear that no sample skipping ever goes on. It's a variable sample rate architecture, whether there is any jitter depends on how the clock input to that CD4024 is generated in the ATTiny2013.

This is very different to an ESQ1.

I have trouble believing you're for real here. Like... what is it about the synthesizer world that gripes you out? Do you hate having more than one synth? Do you want "one synth to rule them all"?

It's clearly not variety you want. You have that. It's not the lack of synths and a wide variety of synth technology available on the market. You have that too.

You want an analog synth with all kinds of morphable waveshapes. So, what, you want a first gen Fairlight or Synclavier which doesn't sample, but generates waveforms electronically? A Prophet~12 or Modulus 002 isn't good enough? You don't want to "cheat" with digital oscillators?

Usually, people demand a solution for which there is no ready product. But the market is already giving you what you want, just not in one instrument. I think you're being a crank about this.

God round up of the Oscar osc's, what else could be said is perhaps the PLL/CTC could be viewed as free run non arbitrating discrete version of a DMA or so, i'm fairly familiar with the Oscar , used to service some of them back in the days, 1985'ish, it's just such a long time ago, don't remember all the details. I had one myself, piggyback Midi version.

Chris is clever yes indeed he is, what i meant by taking it to the edge a bit is the use of frequency doubling by stretching pulses at lest 3 different positions but whatever is enough and works is fine with me!

Do you have any real (measured) figures of the top frequency the PLL ran at?

Anyho,when you made the OSCar emulation did you implement the oscillators slew/rubberband (de)effect? Whenever a user plays a low note and a couple of octaves higher note there is a short slew effect due to the PLL's lock in R/C time constant, known as the rubberband effect if R/C times is to short and slew effect if R/C times to long.

And another thing did you implement the major-oscar-reset-whenever-a-refridgerator-compresor-starts-somtimes-memory-loss design feature?

To be true to the original you must implement both of these there is no way you can cheat by adding regular portamento during pitch change since rubberband is note dependant and the mains spike reset thing i don't now how you actually would implement that into software! But still you simply have to have these in the imPOSCar otherwise it's not 100.00000% genuine emulation!

Currently, if you want to stay more purely in the analogue realm than the likes of the Prophet 12 and Modal 002 (which have digitally generated waveforms, albeit generated in different ways) but want more variety regarding oscillator waveshapes and manipulation, I can see two main options

a) Buy a Schmidt, each oscillator is different and oscillator 4 in particular I don't know of an equivalent to

b) Go Modular, it seems that there's some interesting stuff going on in that area, some inspired by Buchla's designs (or you could buy a Buchla) and some completely original. And of course being modular you can interlink modules to make even more crazy waveshapes.

The Secret World Of Modular Synthesizers

I'm afraid not, I can see why it would happen though. I might have gone on to do it if anyone had asked for it in ten years :-)

I couldn't find a genuine early 1980s fridge to perform my analysis wiith, so I wouldn't have been able to get it quite authentic.
Well since the impOSCar 1 already added polyphony, velocity sensitivity, extra LFOs and extra filter modes (the latter two of which I could actually do on a real OSCar with a firmware change for the former and a little rewiring for the latter), plus of course delay and chorus, and then impOSCar 2 added an extra LFO, unison, stereo spread, polyphonic aftertouch and extended modulation facilities, I like to think of the OSCar as a worthy prototype, but it's understandable if it doesn't perform quite the same as the finished design

Yes it's different from the ESQ1 but you missed the main point in i'm addressing the 002 design difference to Monowave by looking and comparing the 002 PCB with the Defender PCB and Monowave and by what Paul have said so far regarding all these 3 designs, then 002 seams to be quite different then Monowave and a lot more like the Defender. It certainly don't make it easier to figure out when Modal is redefining aliasing and using terms as NCO.They could as easily have said DDS, right!

I assume typo, you meant AT2313.

My take is this, Monowave appears more to be the tradition of a simplified PPG osc using external counters and ROM if you then compare that to Defender and 002 they don't have these externals as far as i can see on the PCB's, combine that what Paul said about having problem with high end fizzle (i recall from the other thread) this suggest Defender/002 is of different design then Monowave add to that i recall Paul once said he was reluctant when something is variable rate or not and since he never replied back regarding high end fizzle in the other thread we cant know... so far.

In the case of Monowave (i recall Paul in a old sdiy thread said they where phase acc's) bit toggling/missing as we know is translated into clock jitter which propagates into the wave read out , but then is LPF filtered by the inclusion of the wave readout counter.

So 002 and Defender with it's non existent counter and ROM parts "appears" to be phase acc's run directly to Flash/SRAM at high speed and assuming 32bits wide with no BLIT is as far as i understand from literature fixed sample rate, why else would Paul have problem with audible high-end fizzle/jitter in the 002? a problem he would not have with a "true" variable rate system.

There seams to be some bugs in the Elby schematics , but the clock you
cant find i suggest is the MSB of frequency generator in the 2313.

You may well be correct, I've yet to see a close up picture of the 002 board or indeed any of the defender, if you know of any then I'd be interested to look at them.

Oops! Good thing we're both better at reading schematics than typing version numbers

One synth to rule them all? where'd you get that? Variety is exactly what I want. I never said I hate what I have or what exists. I want more variety less limitations. I'm not the kind of guy that thinks limitations make for better art.

Dude! I like digital oscillators too! but I recognize analog oscillators sound different than digital ones. I have heard hundreds of both kinds.

Point me to the analog synth that has more waveforms than the basic 4 with only basic FM, square pwm or osc sync to modulate them? Prophet 12 and modulus = digital osc.

Am I saying dig osc sounds bad? am I saying I want analog for analog's sake? No I'm not saying I don't want digital either. I am saying for at least what I've heard analog osc sounds different and I want more variety in them. Telling me what I want exists? great show me. There's a tiny bit it in the microbrute's metalizer and that's why I own one. There are some modular um modules that have further waveshaping options, but modular is expensive and cumbersome and while it's cool and yes I've owned one too, I made it clear from my post that I just want a little more variety from new synths. I think it's kind of rediculous that there's hundreds of synths that really do make the same basic sounds and just some are better than others at certain ones.

And to the guy that recomended a Buchla. Yep it seems to, Thanks for offering to buy me one. The Vermona Perfourmer looks extremely interesting if it can really do 4 oscillator fm and sync! but I doubt the Fm will track pitched across a keyboard in tune.

I was also about to say the Schmidt.

There is nothing particular that stop an engineer form doing complex waves in analogue it's just that it's very expensive to do it. It wont payoff polyphonically so the only place you would find complex waves is in modular/mono dimension.

Holy Schmidt, I must have one! Still don't know what alternate waveform possibilities it has but damn i still want one. Kind of pricey but that's a real instrument right there.

Well, there you go, you found your solution. Hopefully the Schmidt will provide that magic you can't find from anything else.

But, provide some tracks from your search or I'll maintain that you're just being a crank.

I didn't say I found my solution, or that the Schmidt nice as it ma be has much alternate waveform possibilities. And you keep saying things like magic from anything else like nohing is good for me, crank whatever.... The only crank is you, trying to wind me up I also have tons of songs in my sig but I don't care what you maintain, you could maintain I'm a unicorn.

Ya know most people that have innovated have probably been laughed at for daring to attempt something impossible or beyond the limitations of the present time. I just want something that seems like an obvious innovation honestly, and I didn't really think it even was an innovation because I thought the possibility existed and it does somewhat in modulars. From a useage perspective, and not a tradition or feasability perspective it's easy to see why people would want more choice in the oscillator department.

This one seems to be pretty cool in the alternate wave moduluation and fm department, but if the microbrute can have a "metalizer" for under $300 I don't see why other nonmodular synths could have similar but other options.