What is True Analog?

[Dave Bryce]

Okay, here's an interesting question:

 

At what point does an analog synth cease being pure analog?

 

It can be argued (and believe me, it has) that once a synth is under any digital control at all, that it is no longer "pure" - so, is anything with presets eliminated? Are the only true analog synths things like modulars, Minis and 2600s?

 

How about things like a PPG Wave, where the tones are digitally generated, but the filter and amp sections are voltage controlled? Or a Juno 106, which has DCOs - does that make it less of an analog synth than the Juno 60, which had VCOs, if I remember correctly - many people that I know consider DCO synths like the 106 or the Oberheim Matrix-6 to be analog synths...are they wrong?

 

I'll bet that there are some people with opinions on this subject...

 

dB

[Pim]

To me a synth is analogue as long as my ears are telling me it is analogue.

The funny thing is that often the unwanted side effects of analogue synths are characteristic for analogue synths.

The OBX produces for every voice a different sound. Play eight times the same key on an eight voice OBX and you will hear eight different timbres. Even a non-synthesist can hear it. Imperfect. But just because of that it sounds so very alive.

Not so long ago, I sampled some drumsounds into a 20 bit sampler and in the same time thru a TL Audio C1 tube compressor into my 12 bits MPC60.

Everybody liked the "grainy" MPC version more...

 

So when a synth is digital controlled, but with some unwanted side effects (aliasing - PPG), it may have a digital tone generation but our ears will love the sound. Just because it's not perfect.

 

One day digital techniques will be so sofisticated that they can imitate imperfect synths. But that will only happen when synth designers fully understand that perfection is not the goal. The goal is to make a fine musical instrument.

 

 

 

------------------

www.dancewave.nl

[Anderton]

Who cares about purity? Give me a great sound, and you can call it anything you want.

[Dave Bryce]

Originally posted by Anderton:

Who cares about purity? Give me a great sound, and you can call it anything you want.

 

Jeez, ya gotta cut right to the heart of the matter, doncha? http://www.musicplayer.com/ubb/wink.gif

 

I agree with you completely, actually. There are certain virtual analog synths that I'd much rather have than certain real ones...

 

People who know me have heard me say over and over again - I believe that there a five good reasons to buy a synth. They are:

 

1) the sound

2) the sound

3) the sound

4) the feel of the thing (unless it's a module, in which case substitute #3)

5) the sound

 

Follow these rules, and I believe that you are generally assured of satisfaction.

 

dB

 

[This message has been edited by Dave Bryce (edited 10-15-2000).]

[Erik Norlander]

Craig and Dave are spot on about the sound being the ultimate point. Once these analog modeling synths reach the level of sophistication where they have all the character, life and ever-evolving unpredictable originality of an actual analog machine, I'll be the first to sign up to them. Right now, though, the difference is still too great for me. The Moogs, Oberheims and Roland MKS modules stay in the studio...

 

The analog modeling synths are wonderful technological achievements, though, so by all means, you engineers, keep working! One day this technology will likely get there.

 

Cheers,

 

Erik (who, although a skeptic of analog modeling synths, is an avid enthusiast of the Yamaha VL physical modeling synths)

[Guest]

Certain types of artifacts (such as stairstepping when you crank the filter cutoff knob) characterize a hybrid instrument. Such artifacts may even be musically desirable at times, but they're a problem (and rightly so) for the analog purist.

 

It's not terribly difficult to build a digital "analog modeled" synth that avoids these artifacts, I believe. It's just more expensive.

 

Another area where digital synths tend not to respond like their analog brethren is in the extreme high and low registers. Highs can produce aliasing, for instance, which you won't find on an analog synth.

 

Having said all that, IMHO only a masochist would insist on a true all-digital instrument. I've worked a _lot_ with patch cords on a modular, and as inspiring as the sound can be, getting anything done musically is a nightmare -- unless you have a nice digital recorder handy to capture the sounds. Even then, tuning is likely to be a problem.

 

In the end, I'm willing to sacrifice a certain amount of sound quality for the sake of convenience. But then, my music is more in the trad pop line. It has chord progressions and melodies and stuff. YMMV.

 

--JA

[Erik Norlander]

Jim, you wrote:

 

"IMHO only a masochist would insist on a true all-digital instrument. "

 

But I think you meant "all-analog instrument", right?

 

While I understand your point, I don't think these analog instruments are all that brutal. Sure, microprocessor-enabled recall ability is great, but if you dial up (or patch together...) the same kinds of sounds repeatedly, it simply becomes a part of the music making process just like tuning a guitar or setting up a drumkit. Yes, there's some slop involved, and true, you will never get *exactly* the same sound twice, but ooh, vive le differance! (Sorry, I don't know exactly how to spell this French expression.)

 

All that said, I'm still holding onto my samplers and digital synths. For overly complex multi-layered tapestries, total recall is definitely essential. Everything has its place, eh?

 

Cheers,

 

Erik

[Guest]

All-analog. Yes, sorry am very being. Editor's brain fried yesterday was.

 

Comparing patching an analog modular to tuning a guitar is a bit of a stretch. Tuning a guitar shouldn't take more than 30 seconds, tops. For setting up a working patch on a 4-panel Serge Modular, including all of the little dial-tweaks, count on 10 minutes minimum. (Plus the time you spent writing down the patch last time, before you disassembled it.)

 

--JA

[Oli P]

To me it was a really big change when the DCO synths came....

I really had a hard time getting them to blend in big arrangements, like I was used to the Jupiter did. It is like the tone of a DCO is very narrow, and contains a lot less overtones than a real VCO.

The Jupiter6 was digitally controlled on the filters and such, but still had VCO's ...I liked that approach a lot better.

So...to me a real analog synth has real VCO's.

 

I still dig the Virus and Z1...mind you http://www.musicplayer.com/ubb/smile.gif

[Erik Norlander]

Originally posted by Oli P:

To me it was a really big change when the DCO synths came....

I really had a hard time getting them to blend in big arrangements, like I was used to the Jupiter did. It is like the tone of a DCO is very narrow, and contains a lot less overtones than a real VCO.

The Jupiter6 was digitally controlled on the filters and such, but still had VCO's ...I liked that approach a lot better.

So...to me a real analog synth has real VCO's.

 

I think the best part about analog synths is that they drift and roll around. DCOs don't move nearly as much, and that kinda defeats the purpose for me. For fatness, life and general organic qualities, I prefer the VCO sound. Mike Peake, make sure that Andromeda is good and drifty!

 

Originally posted by Oli P:

I still dig the Virus and Z1...mind you http://www.musicplayer.com/ubb/smile.gif

 

For *some* applications, though, DCO machines are nice. I personally enjoy my Roland MKS-70 for sync stuff and controlled pads and horn blasts. Often I prefer the DCO MKS-70 to the VCO MKS-80, surprisingly. Although the MKS-70 filters do have a certain unique quality, too. And I love that era's Roland chorus effect.

 

Okay, so can I hear an MKS-70 with MKS-80 oscillators? Anybody? http://www.musicplayer.com/ubb/wink.gif

 

Cheers,

 

Erik

[Jim Aikin]

Oscillator drift is one of those quasi-mythical phenomena. Anybody ever _hear_ an oscillator drift? I have. The oscillators on my old Serge Modular used to take 15 or 20 minutes to warm up, but that simply meant you couldn't play any sort of in-tune music during that period. The sound quality didn't change appreciably between the warmup and post-warmup periods. The beating in 2-oscillator tones would change rate slowly.

 

It may be true (I'm not an engineer) that you can't get a phase-lock between two analog oscillators, but so what?

 

The real differences between analog and digital oscillator (and filter) tone probably have more to do with the shape of the waveforms. Here's a challenge, if anybody is up for it. Dust off your Prophet-5 and sample a few tones into your favorite digital audio recorder: One oscillator at various pitches, with and without filtering. Now do the same with, say, a Supernova, AN1x, or Q. A/B the waveform displays, note the differences, and try resampling with various variables moved around to try to get the waveforms to be identical. I'll be interested to hear/read/view the results.

 

Heck, I'll even read them if they're in a competing magazine. I'd love to do the test myself, but my plate is kinda full this month. I'm happy to let the guys in the lab coats do it. !-o

 

--Jim A.

[marino]

Well, I tried to sample my Rhodes Chroma several times, especially right before selling it. (the stupidest move of my whole life)

I tried around '93 with a Roland S-550, then with my K2000RS. I was very careful to keep the cleanest and hotter signal path possible. Soon I resigned myself to sample only those sounds with a certain static quality. I patiently reworked the envelopes and filters on the Kurzweil. The results were ALWAYS very poor.

I said, maybe I'm having audio hallucinations, just because the sound is coming from a different box, I'm hearing it differently. So I started making comparative recordings. Even blindfold tests with friends. NO WAY.

The point is: the analog sound (that particular analog sound, at least) moves. Every time you strike a key, it's a little different. And the sampled sound is.... smaller. The beating beetween oscillators is constantly changing, and you don't get that in the sample.

Then of course there is the problem of real time controllers.... I think that sampling a grand piano is much easier than sampling an analog synth. I know that after a few years I gave up completely.

[Dave Bryce]

Originally posted by marino:

The point is: the analog sound (that particular analog sound, at least) moves. Every time you strike a key, it's a little different. And the sampled sound is.... smaller. The beating beetween oscillators is constantly changing, and you don't get that in the sample.

 

Fully agreed. The voltage controlled stuff drifts, moves, shifts and changes constantly. I've never heard a MiniMoog sample that I thought did the instrument justice - it's just not possible (IMO) to capture the life of the thing by taking a sonic snapshot of it. Don't get me wrong - you can get some very nice, very fat sounds...

 

...but it ain't the same.

 

dB

[synthfoolsynthfool.com]

I used to think that an oscillator had only the following attributes to it:

 

Frequency

Waveform

Amplitude

Phase

 

But now Im certain that there are also factors relating to how much variance there are in these attributes as well.

 

For frequency, there have always been wow and flutter, a term used mostly for frequency and amplitude variations in record and tape mediums, but still a measurable attribute with VCOs.

But there can also be CV / Freq overshoot, ringing, waveforms that lose or gain amplitude with frequency, phase changes, output loading and self synchonization and power supply filtering and decoupling issues.

Lotsa technical forks stuck in there.

 

Its probably possible to take perfect oscillators and by injecting some sort of low frequency noise or other randomness, induce a more sonorous tone.

I just havent seen any perfect oscillators... ;-)

 

Hearing the difference between a single Moog 901B and a CEM3340 (or DCO) is pretty obvious to me. Two of them beating, even moreso.

 

 

K

[Peake]

Sampling is indeed like taking a few frames from a movie and then missing what's left on the cutting room floor. And anything in the audio chain (including a sampler) changes the sound.

 

And Kevin and Erik are (of course http://www.musicplayer.com/ubb/smile.gif right: there are very subtle characteristics that cause one oscillator to sound more musical than another (within personal preference). Erik just found that the polarity of a sawtooth is audible; I found that a year or so ago, in my preference for the Moog 901b saw over the 921b saw. It turns out that they had different polarities, and the ear can prefer one over the other. Add a full spectrum of harmonics and let two of them drift against each other and it's music...

 

Jim, I've looked at some waveforms of synths before. The EMS Putney has a slightly different saw overshoot (high harmonic content) upon every wavecycle. I doubt that this sort of thing has been modeled yet, though Waldorf has actually mimic'd the circuitry of an oscillator instead of just generating a waveform... The original Nord Lead actually has a decent sawtooth, if you listen to one at a time. It's in the drift and beating that it begins to lose interest, and then the filter.. Well, I still have to hear a Lead 3, don't I http://www.musicplayer.com/ubb/smile.gif

 

But for me, a synth stops being analog when you can't patch it by hand. It's not a measure of value or musical dimension, it's about serendipity and potential. And as Erik and Jim have experienced, when you need repeatability, computer control counts rather a bit http://www.musicplayer.com/ubb/smile.gif

[maxmallenyahoo.com]

The whole DCO vs VCO thing is very interesting/confusing.

Mike - would you care to shed some light on the subject/

 

tia

 

Michael Allen BALD guy

[Peake]

Mr. Allen! Nice to see you here.

 

This just came up on Analogue Heaven, so I'll attempt to share what I learned. Search the AH archives for DCO for more info: http://www.midiwall.com/aharchives

 

A DCO is a Digitally Controlled Oscillator. Typically, it's an analog oscillator that is hard sync'd to a digitally-generated tone, essentially creating a perfectly-stable tone with a bit of the characteristic of analog. The "perfectly-stable" aspect is what puts me off, as the natural analog drift is what makes it warmer and more interesting to my ears. Also, a hard-synchronized oscillator never sounds as fat as a free-running oscillator, due to aspects of the synchronization results and even worse, if the digital tone source it is sync'd to always resets it's phase upon each new note (this causes any detuning between a pair of oscillators to always start at the same point and moves out from there. It's quite different from the continuous beat pattern generated by a pair of unsynchronized (free-running) oscillators, such as in the Minimoog etc.

 

Hard Sync is an aspect of many analog oscillators that causes the oscillator to reset it's phase upon receiving the rising edge of another oscillator's waveform. In unison with hard sync on, the sync'd oscillator will have the exact same phase as the master, and will cause a volume increase as a result (addition). If a hard sync'd oscillator is higher in pitch than the master, it's phase will still reset upon receiving the rising edge of the master's waveform, but will then generate as many cycles as it's pitch is set to, and will then reset upon receiving the next rising edge of the master oscillator's waveform, etc. All of these in-between cycles that are not in phase lock with the master are the reason that a hard sync'd oscillator produces harmonics of the master, and loses it's own waveform. There is a better technical explanation I'm certain, but that's the gist.

Zon might have a better explanation?

 

BTW, some synths are simply digital waveform generators, such as IIRC the Juno 106, etc.

 

Check the AH archives. A few folks who know more than I do spelled it out quite well.

[synthfoolsynthfool.com]

Not to find fault with Mike's explanation of DCO's, but by far most DCOs are actually not analog oscillators synced to a digital sources, but usually are made by devices known as "programmable counters".

Almost all of Roland's DCO synths use this scheme.

 

A programmable counter is simply a digital counter IC that is set to cycle (loop). Give them a binary number and they count up to the value set by that number and then start over again.

The waveforms are usually directly related (square / pulse) or formed by addtional waveform converters later.

 

Mike is completely correct however about why DCO are's generally not too interesting- they all start with the same (usually crystal based) clock source and thus the phase relationships between DCOs remain constant.

Even though various analog techniques are used afterwards in waveform modification, the original waveform is locked to the others in phase and is very hard to change later without separate phase shift type circuits.

 

Hope this helps explain things without stepping on any toes http://www.musicplayer.com/ubb/smile.gif

[Peake]

Ah! Thanks Kevin. I'll bet that I remembered the -incorrect- definition of DCO from the AH discussion, and posted it here. I knew that I was either wrong or potentially incorrect enough to justify a recommendation for people to check the archives etc.

 

This is why discussions often consist of more than one person

 

I suppose I don't know about DCOs because for the most part, I find them completely uninteresting! (I don't retain much that doesn't interest me, and I'm not afraid to admit it

 

There -is- interest in synths like this (to me, purely digital tone sources such as the PPG through true analog filters, even samples through analog filters ala' the Emulator 3, are interesting to highly interesting). However, for "true analog", you have to stand by an entirely analog signal path at the minimum. DCOs don't sound like analog oscillators. Digital control of analog signal paths doesn't change the sound. For -that- sound, true analog is it. And then there are the different levels of analog... the 'pinched', high-harmonic distortion of the Curtis chips versus the much better sounding SSM chips, to the differing levels of op-amp based and transistor based circuitry. My ears prefer the earliest, most discrete designs (and the price tags on this gear reflects the tone quality). Even with the potential for oscillator drift etc., there was a musicality and electricity to the earliest gear that the curtis chips IMO seemed to misplace (and which sadly replaced the sonically superior but static-sensitive TEMPCO-resistor hungry SSM chips...)

[maxmallenyahoo.com]

Thanks for the info guys http://www.musicplayer.com/ubb/biggrin.gif

 

Michael Allen BALD Guy

[ITGITC]

http://www.sonicftp.com/news/images/dsi_mek.jpg

 

It's now 2007.

 

What about the MEK? How does it hold up in the scheme of things seven years later?

 

[MikeNZ]

I think that part of it is the cynical attempt of the makers to try to recreate analog synth sounds in cheap keyboards. The old Moog, Prophet and others were high quality instruments, designed to be played by discerning musicians. Compare that to a $400 alesis Micron - with the cheapest possible keyboard, case, and controls. The fact is that instruments like the new Moogs and the Dave Smiths are EXPENSIVE as you have to pay real money to get both the sound and the interface to make that sound work as an expressive instrument.

 

Digital has given us huge polyphony, huge sound banks, and multi-timbral sounds. We want our sounds for pennies per sound, and dollars per note of polyphony. Why pay $1,200 for a keyboard instrument that can only play one note! We have (to some extenet) traded the soul of music for the cheap trinkets of digital.

 

We are spoilt with modern keyboards and to some extent have forgotton what playing a MUSICAL INSTRUMENT is about, which is the intimate connection being the musician and the notes (music) which is being played, to create an intimate connection with other musicians and with the listener.

 

Most ROMPLERS (especially cheap ones) have no feeling of intimacy. My Korg N5ex certainly didn't (as good as it sounded), but my PC1X does on some sounds.

 

Michael

[RABid]

I guess I missed this one the first time. For ME, true analog is being analog from one of the signal path to the other. Osc's, filters, effects, etc. The only controller I might be concerned about is pitch bend.

 

Something like the PEK I would classify as "close enough".

 

Robert

[mildbill]

Well, I'm a little confused by Synthfool's definition.

 

I always thought, (and still do), that a 'DCO' was an analog oscillator under digital control.

 

What he describes is what I would consider to be a 'digital oscillator', or a digitally generated oscillator, or a software oscillator.

 

Anyways, 'What is True Analog?'

 

I think if it's got a VCO, VCF, and VCA, it's true analog.

[KenElevenShadows]

Okay, here's an interesting question:

 

At what point does an analog synth cease being pure analog?

 

When a synth stops drifting in pitch or tone.

[Jim Alfredson]

Don't ask me, I just play The Beast. I believe it is true analog, though...

[Dave Bryce]

I always thought, (and still do), that a 'DCO' was an analog oscillator under digital control.

Me, too. The ones in the Evos certainly are.

 

What he describes is what I would consider to be a 'digital oscillator', or a digitally generated oscillator, or a software oscillator.

I agree, and I'm not sure Kevin is right...Dave Smith would say he isn't, for sure...but Kevin Lightner (synthfool) is an incredible synth repair tech, so maybe he is right and maybe Roland's DCOs work differently than most others.

 

I'm way curious now...I'll make some inquiries.

 

dB

[mate stubb]

Don't ask me, I just play The Beast. I believe it is true analog, though...

 

Not just analogue, but electromechanical - a higher state of analogueness!

[The Real MC]

If the audio signal path is 100% analog with no transformation into the digital domain then it should classify as an analog synth, but you should divide analog synths into continuous analog and discrete analog.

 

Minimoogs, 2600, modulars, et al would be continuous analog synths as their control voltages are continuously variable with no discrete stepping from the digital domain. Sweeping the filter is completely smooth, a desireable attribute in some musical styles. The downside is if you designed a patch storage system for a continuous analog synth you would not get storage for ALL the parameters. Witness the Oberheim SEM Polysynth programmer which cannot store filter resonance and other parameters.

 

Discrete analog synths derive most or all control voltages from the digital domain and there will be audible stepping (with some non-audible exceptions). The Prophet-5 has a 100% audio signal path but the stepping is very audible when you sweep the filter from the front panel. That stepping is not ideal in some musical styles but this system is ideal for patch storage.

 

Some folks seek to split hairs whether the EGs are analog or digital. Being the analog snob I am, it does not matter either way. You can design a digital EG and the stepping will be inaudible. Consider the Moog Source. The EGs are a voltage controlled integrator whose transient times are set by control voltages from the DAC. So even though the EG transient times are set by discrete control voltages, the resulting envelope is pure analog with no stepping. The Voyager has the same design. CEM3310 EGs are a similar design. The Andromeda EGs are software generated and I hear no stepping on slow transients.

[Jeff Klopmeyer]

Jesus, Gas: you did a 6-1/2 year thread dredge? Was the water murky there at the bottom of the forum pond?

 

Jeff sez: analog is only about audio signal path, no matter what type of equipment we're talking about. If a synth has analog oscillators and the sgnal from those can go unencumbered through to the outputs, it's an analog synth. Anything else beyond that gets fuzzy, but it's only interesting if you like talking about semantics.