Saturday, February 20, 2010

Tim Smith on Analog Synths versus Digital Synths

Entered as found except for subtle correction; all italics and bold are in the original.

October of 93

Why do analog synthesizers sound so different from their digital counterparts?

This is a subject I delved into with some detail in Mark Vail's book 'Vintage Synthesizers'. (Available through Keyboard Miller-Freeman Publications.) Mark made an excellent contribution to the history of electronic music by compiling this text. I peruse it often. In the interview with Mark, I explained some of the basics of audio electronics. I'll try to present a comprehensible overview of this subject again here. Let's look at the Moog- mini Moog. When this instrument was invented, designing with IC op amps was in its infancy. There were very few available, and if one was any good, it was really a monolithic device constructed out of discrete transistors, resistors, capacitors, etc. Maybe encapsulated in epoxy, etc. The device that was available commonly was the UA741 op amp. The UA741 device was used in the mini-moog, but it was used only in the Logarithmic amplifier circuits that convert the linear control voltage from the keyboard and other sources into an exponential current suitable to produce the doubling of the sinusoid's response necessary to produce a musical octave. (This current was used to control the charging current in the integrator that is the basis of the oscillator in the mini). As a result of the lack of IC op amps, the mini design team designed almost everything with discrete circuit configurations, much like high end audio designers would have done in hifi gear. The result? A very hifi sounding synthesizer!

The point that I'm trying to get around to is this: The mini-moog had no IC op amps anywhere in the audio signal path. Thank goodness, or I'm certain that it would not be teh legendary instrument that it is today. I think it's important to point out that the Moog modular instrument's system I, II and III had no ICs in the signal path either. The Moog System 15, 35 and 55 had very few op amps in the signal path, I think around 1. (Edit: IIRC there are opamps in the outputs of each oscillator waveform and also one in the CP3a mixer). If you had the good fortune of reading Mark Vail's article about Emerson's Moog system, you may have noted his reference to changing the op amps in this system and the very positive effect it had on Emerson's Moog! (They used National's LF411, 412 Types, a quiet, stable op amp of more contemporary design). The 741 type op amp designs would have ruined the sparkling, sweet highs that made the mini and the early modular instruments so famous. The 741 is not suitable for use above a volt or so about 6KHZ, so you can see why.

Op amps from this period and even some modern ones act like low pass filters in an audio circuit, or worse, when pushed to their limits, generating slewing induced distortion- a very unmusical sound, some folks describe as a smearing of the signal. Until recently the best Pro audio and HiFi equipment did not use ICs either. That's right, in 1993! Some engineers will probably never accept op amps in the IC form from a purist's point of view. There is not a small number of poeple that are convinced that Vacuum tubes are the finest form of sonic amplification and they certainly have a point. However tubes are of little practical value in Voltage controlled synthesis circuits.

The lack of acceptance for IC op amps by engineers and careful listeners in the High Fidelity and high end audio fields is not without reason. Until recently, thay have not had very good sonics, or if they had a nice warm sound they were somewhat noisy. Thankfully this is no longer the case. But unfortunately, even today's great digital synth companies are not using the best op amps they could. There are other factors in the mini-moog that are important, like the Moog lader VCF, who would dispute this? Also the frequency response of the mini is more or less flat out to 35KHz. I recently fired up my Arp2601/3620/2500 system and was re-shocked at the fat, rich sound that comes out of this system. It has two 2500 series 1407 multimode filters and 1 of the early Arp copies of the Moog VCF installed and a total of 11 VCOs. It's 2 note duophonic but I usually just use the monophonic mode. What a note! My poor sample playback machines are somewhat sad in comparison, but certainly have their own merits. I've upgraded all the op amps in my analog systems (I have 3 souped up 2600s along with the 2500 system). It made a very real difference! Remember; old, slow op amps are Low Pass Filters.

So what's the difference between old analogs and new analogs, or old analogs and new digital instruments? Let's start with old analogs versus new analogs, or particulary the polyphonic analogs like the Prophet 5, ESQ-1, the Oberheims with Curtis ICs in them. They are all constructed with IC versions of the synthesizer module, VCOs, VCFs, VCAs etc. The most prolific are the Curtis ICs. I can tell you for certain that they are sonically much inferior to the early synths, particulary modular synthesizers. I have had scores of folks tell me the same thing!

Why? These IC synth functions do not have the simple, unobtrusive signal path that the modular machines did. They have relatively high THD, usually more than 1% (odd order unfortunately) and they have what I usually descrive as a flat, 2-dimensional, pinched sound. So what happened? In the change from discrete "hifi" type circuits to the IC form, something was lost. The early Moog and ARP VCAs (2500-2600-900 series Moog) had discrete design with greater than 90dB of dynamic range. Digital before Digital!

What changed when we went to digital, relative to analog? If you read Mark Vail's article about Emerson's Moog, you may recall one of the technicians who modified this instrument mentioning that the Moog's response was extending out to 50KHz or so. All digital synths have a BRICK WALL filter at exactly 20KHz. Is there a sonic difference to these two systems? I sure hear one! Let's face it, a Moog 55 can demand nearly every dollar it sold for on the day of its manufacture. Now that's lasting value. It isn't just because it's a pretty synthesizer either! (Although it certainly is!) ARPs in excellent shape can demand the same arrangement, especially the 2500 system.

I'm not saying that the frequency response as a singular spec is the whole answer; it's not. I believe that a major factor is this; In high end hifi design schools, simple topologies (circuit layouts) are widely believed to have the most desirable sonics. The less nonlinear nodes (connection points) that a signal has to go through, all the better. Hence Vacuum tube designs with their very straightforward topologies, and next best- discrete FET designs, then Bipolar transistor designs, and the least desirable topology? ICs (integrated circuits). But I must say, today's op amps are changing minds and the rules. There are some really excellent-sounding devices around now, as good as the best discrete designs in my opinion. Let us install some of these marvels in your system!

Can our mods make your digital synth sound more analog? Well, sort of. We are certainly speeding up the response time of these circuits (slew rate, settling times, etc.) . We are also electively raising the frequency response in the Reconstruction Filters, and if there are electrolytic capacitor in the signal path that aren't rerquired and are causing unnecessary low freq. roll offs and phase shifts, we can remove them! If they are required, we can correct their unpleasant effects. So at least these machines are performing at their very best. These mods are standard fare in the HiFi mod world. We are just bringing them into the Pro Audio, electronic music world.

Will these mods make your Vintage Keys (or other type) module sound like a Moog modular or an ARP modular? (I love mine, it sounded quite nice before new op amps, and even smoother and warmer after new op amps were installed). This would be an extreme claim to make, but let's say this: The sound of a digital synth with the best possible op amps and if necessary an extension of its frequency response certainly gets you closer! I think a lot closer. And our customers agree! The lastest audio op amps are discrete, monolithic designs. It's a good age for op amps, they are finally up to snuff! When you hear or read about people saying their old analogs are much warmer and fatter sounding, they are not just hearing things (no pun intended). They are REALLY HEARING THINGS!

Another point I would like to make is that almost every digital synteh has a very sharp (24-30dB per octave) filter set at usually no higher than 20KHz. There are studies that show while we may 'not hear' abouve 20KHz, there is EEG activity above this frequency in certain folks. There is a perception of energy above this point. Plus if you will look at the gain/phase graphs (showing the changes we make to the RCF Filter) I have provided in this brochure, you will see significant phase shifts well within the Audio Bandpass. If you extend the frequency response of a digital synth's circuits you move this phase curve further away from the ear's perception. This is helpful. Well, if you're still awake and not entirely confused, I hope this has given you something to chew on. Thanks for taking the time to read my dribblings.

Timothy C. Smith

Weyer/Smith Labs

The Audio Clinic