How fast is fast?

The Moog Forum Minimoog Forum
1

OK the text below is from an SOS review of the Technosaurus Modular:

The Minimoog is revered, in part, because of the speed of its envelope generators. Although its front panel shows a fastest attack time of 10mS, common belief suggests that a Minimoog responds about 10 times faster. Now hold on to your hats…

Again using the S770 as a diagnostic tool, I set a Selector ADSR to 0,0,0,0 and used this as the CV for one of the VCAs. I then fired off a single trigger, and inspected the resulting waveform. I repeated the process until I believed what I was seeing. The rise time from the trigger to the maximum amplitude was typically eight samples. At a sample rate of 48kHz that’s just 1/6000th of a second, or 167µS (167 microseconds). The fastest result I obtained was a blindingly fast five samples, which is only 104µS and consistent with Technosaurus’s claim of 70µS for a 90-percent response at the VCA’s output.

Since I still couldn’t believe this, I set up my Minimoog and performed the same test. Typical attacks lasted about 30 samples (625µS) which was much better than I had expected, but six times slower than the Selector! These results are incredible. It makes the Technosaurus far snappier than any other analogue you’ve ever heard, and adds an impact and presence that no other analogue synth can emulate.

Oops… amidst all this excitement, I’ve forgotten to tell you the envelope generators’ specifications. The six EGs come as three dual devices, each of which offers DADSR (delay/attack/decay/sustain/release) and DAD (delay/attack/decay) profiles, with inversion if required. We’ve covered the fastest attack times in some depth, but it’s also worth noting that the fastest decay and release times are also in the sub-millisecond range. At the other extreme, the slowest times are around 10S, which is sufficient for most purposes.

Does anyone know how fast are the envelopes of the Voyager?

2

The Voyager envelopes have a minimum 2ms attack transient. The vintage minimoog is 2ms, while most vintage analogs can’t match that.

The other component in the equation is the VCA. Slamming a VCA with a microsecond transient is one thing. Getting the VCA to perform with minimum control voltage feedthrough is another, which manifests itself as a nasty THUMP on the audio output. The faster the transient, the more likely that feedthrough will creep up.

3

But, if you had to pull out a tool to tell the difference vs using your ears, then there’s no audible difference- so what’s the point?

4

Hello

The Voyager envelopes have a minimum 2ms attack transient. The vintage minimoog is 2ms, while most vintage analogs can’t match that.

Yeah, this is true but I have noticed that most synthesizers start with a 2ms attack. This makes sense in digital envelopes since enveloping a signal is nothing but multiplication of signals, in this case multiplication of the VCA output/level with the envelope. If the envelope started at 0, then 0 * whatever = 0 which means there would be no output. If it was 1, then 1 * whatever = whatever so I suppose the VCA would be always at a steady level?

By the way, I have heard that the envelopes of the original Minimoog clip lightly and maybe for a few ms which is what accounts for the thump in some of the bass sounds (yet another trick?!). I have tried this in the Andromeda by messing around with the predelay and attack times and offsetting the envelope and it can cause a subtle difference.

By the way, MC, I was looking at the oscilloscopes pictures of yours, where you did a comparison between the Mini and the Voyager. Very interesting to see these… and if I am not mistaken is there a subtle (again) difference in the output level of the waveforms?

But, if you had to pull out a tool to tell the difference vs using your ears, then there’s no audible difference- so what’s the point?

I suppose there is… many synths claim 2 ms attack but is it really the case? Some envelopes exhibit a click with 2ms attacks. I am not an EE, I have been told that this is the price you have to pay for very fast envelopes, but in some other synths the click is absent, haven’t you noticed this?

Yannis.

5

There are three main reasons why the output of a VCA will click when triggered by a fast attack control voltage [CV].

  1. The dreaded CV breakthrough. Asymmetry in the VCA circuitry and PCB layout will cause the CV to leak into the audio line. What you are hearing is a smaller version of the ADSR CV itself, or a differentiated version of it if it is capacitively coupled into the audio line.

  2. DC offset on the input to the VCA. If the waveform going into the VCA does not have a mean value of zero volts you will hear a thump. This is present on many synths that use a DC coupled input stage to the VCA - the ARP2600 comes to mind and is easily solved with a DC blocking cap at the input.

  3. The most misunderstood one. Any continuous waveform that is suddenly allowed to change volume will produce a click. This is most apparent on low frequency sine waves or other low harmonic sounds. If the waveform is suddenly turned on by the fast attack [or switched off by a sudden release] then the output of the VCA rises [or falls] very fast producing a sharp edge. This discontinuity is heard by the ear as a click.

You can get around the click by making sure the waveform is at zero volts when you turn it on. This is hard to do in an analogue synth, although samplers do it all the time. Or you select a suitably slow attack time to ensure the waveform, wherever it starts, is allowed to rise gracefully. The higher the frequency the faster the attack time needs to be hear the click.

In general bass notes with few harmonics need less of a fast attack, otherwise they will click.

6

Any continuous waveform that is suddenly allowed to change volume will produce a click

You imply that there exist non-continuous waveforms?

My take is that with very fast attack times, what happens is similar to oscillator synchronization where a transient forces the waveforms to restart - that transient produces additional harmonics which were not present in the original waveform. Probably this happens to all waveforms, it is indeed most noticeable to low-frequency waveforms. Sawtooth and pulse probably hide it better in the attack phase because there exist a lot of harmonics.

I am not sure if this is case of discontinuity, if you set any decent analog to adsr = 0,0,0,0 and sequence anything you will hear a small series of clicks - does that imply that all exhibit discontinuities?

DC doesn’t necessarily causes clicks - at least not in the A6, for instance.

Y.

7

Actually, all waveforms are non continuous by the fact that they do start and stop at some point in time. Nothing lasts forever. But for most purposes in audio, if the waveform is constant over several seconds, it can probably be construed as continuous.

But non continuous waveforms are things that have no specific frequency, like a transient spike on the power supply, or indeed the CV from an ADSR.

My take is that with very fast attack times, what happens is similar to oscillator synchronization where a transient forces the waveforms to restart - that transient produces additional harmonics which were not present in the original waveform.

Yes, a similar event is happening here. Creating a sharp change in output voltage. But its not quite the same thing since all we are doing in the VCA is turning the waveform on and off. Turning a signal rapidly on after it has been off will produce harmonics that last for a short period of time - a click.

Interestingly, some designs of VCOs will sync the output to zero volts. This is particularly useful for some LFO applications. I think the Moog 921 allows variable sync level.

I am not sure if this is case of discontinuity, if you set any decent analog to adsr = 0,0,0,0 and sequence anything you will hear a small series of clicks - does that imply that all exhibit discontinuities?

A perfect VCA being fed by an ADSR set to produce a transient pulse will not produce a click with no input.

With an audio input, you will gets clicks. Each click representing a small clip of the original signal. That clip will be shaped by the ADSR waveform, and will have envelope rise and decay time of the order of 1 to 2mS. The clicks you hear are both due the enveloped audio signal and the fact that it has been chopped into sharp edged packets.

But try this, use a single sine waveform only of low frequency. Then listen as the clicks fade in and out. This is because sometimes the trigger that controls the ADSR will hit different parts of the input waveform. Where the VCA is opened at a zero crossing, you will hear only a small click. But at the peaks of the sinewave, you’ll get a louder click.

DC doesn’t necessarily causes clicks - at least not in the A6, for instance.

It will always click since the output of the VCA has risen rapidly from nothing to something. This discontinuity will be heard as a click. But, whether DC gets through the filter and hence to the VCA is unlikely. Most VCAs and filters are not DC coupled and thus will filter out any constant DC.

8

Actually, all waveforms are non continuous by the fact that they do start and stop at some point in time

Sure they are continuous, but not because they start and stop at some point in time but because they are pereceived as such. If they were discrete they would also have a minimal duration.

But non continuous waveforms are things that have no specific frequency,

which contradicts with what you wrote before. Just because it doesn’t have a specific frequency it doesn’t mean that it is not continuous. How about noise? Noise is supposed to have all frequencies but it cannot be regarded as discrete, can it?

With an audio input, you will gets clicks. Each click representing a small clip of the original signal. That clip will be shaped by the ADSR waveform, and will have envelope rise and decay time of the order of 1 to 2mS. The clicks you hear are both due the enveloped audio signal and the fact that it has been chopped into sharp edged packets

If we could hear 2ms of the original waveform, without enveloping the signal, it would certainly be different. I don’t see what you say it is a small “clip” of the original waveform. If it clips, it will probably because the transient introduces the extra harmonics for a very short duration, then it shuts and the ear cannot perceive the waveform as such. Methinks that it could also be some form of modulation since we are talking analog envelopes here, but I am not sure.

Y.