Roland Alpha Juno-2 pwm sawtooth

[itoa]

I'm trying to reconstruct the Alpha Juno 2 saw pwm oscillator.

Initially it looks like a saw wave ring modulated by octave higher pulse. But not quite

Any tips?

My observations:

- Segments have variable steepness, depending on pulse width. Lower part looks like it's low pass filtered - nonlinear filtering?
- Initial saw is rounded - this could be effect of low shelving or ring modulated sine - but it's not, it ends up as a straight line.
- Maybe 'integrated signal' in DCO is stopped during negative pulses?

[2DaT]

Looks like dc-blocking highpass filtering.

[itoa]

Not quite... There is DC blocker but its contribution to the shape is minor. The roundings are present at higher freqs as well.

[MadBrain]

I remember reading somewhere that the juno pulse looked something like this:

Code: Select all

________________________________|¯¯¯¯|____|¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯

(with the ratio between the short and long segments varying by pulse width)

This might not be accurate.

[itoa]

Not related here, but very interesting information. Waldorf Rocket has the same - symmetrical pulse. PW gets inverted (1-prevPW) every cycle.

[mystran]

Not quite... There is DC blocker but its contribution to the shape is minor. The roundings are present at higher freqs as well.

Probably a high-pass filter that tracks the fundamental .. such a thing can be used with naive waveforms to filter out any aliased partials below the fundamental.

Btw, these plots are time-reversed, right? Is there a reason for that?

[itoa]

- No the time is not reversed - and this is the problem :]. - this is a rounded rising sawtooth. Maybe the wave shaping is a key.
- Some parts look like hp filtered, some like lp filtered. I've tried different combinations of filters - tracking and not - before or after modulation.
- Can't find schemes of Alpha DCO. AFAIK 'digital signal' from cpu loads capacitor and this can be abused here . So far is a total black-box to me.

Link to the original recording:
https://dl.dropboxusercontent.com/u/117 ... %20pwm.wav

btw. Just realized Audiorealism Redominator does the job perfectly

[Sendy]

The octave higher pulse is amplitude modulated, not ring modulated. When the pulse is high, the original sawtooth is let through; when it's low, the waveform goes low. Sometimes this is called "combing" a waveform as you're inserting regular holes in it (not to be confused with combing in the spectral domain, i.e. with a comb filter). The lines are straight and hence it's easier to see without the highpass filtering.

You can make this family of waveforms with any synth with sync (to phase lock the oscillators), two oscillators, ring modulation and a mixer (to make AM out of RM).

EDIT: May as well trot this out again



These waves were made in the freeware TAL Noisemaker, but any decent synth will do the job. DIVA also emulates these waves with it's DCO oscillator section. (Well, at least, the combed saw, a la Alpha Juno)

[itoa]

Sendy, the devil lies in the details and I actually asked about the devil

I started from the method you described but if you look at the original Juno it looks like RM modulated - and in fact it sounds closer than AM (look at the curves).

Both methods are just rough approximations.. Juno sounds different, more edgy, first partials are more static. Just compare the waveforms, lower pulse is sloped, higher remains the same except the last part .

e.g. I've been trying to achieve the falling slope of the first pulse using HP filtering.. Got this but upper pulses were altered as well.

I suspect it's not about a trivial filtering of the final product

[aciddose]

The wave is a plain up-ramp AND a pwm at (edit: twice the) frequency. You can also use any frequency you like, 2, 3, 4 octaves for example to get more variations.

The narrow pulses are as far as I remember 3 octaves up, that is 8 pulses per ramp. They're perfectly square.

In terms of "AND", we're talking not bitwise but analog. That is AND just means to amplitude modulate the ramp by the pulse, where the pulse is between 0 and 1.

By the way, to avoid this problem in the future sample the waveforms at a higher sample rate and play a much higher frequency. You'll then see the original unfiltered waveform shape rather than having all the low frequency harmonics removed.

For example, sample at 96khz and play a C6 at ~1000hz.

[Sendy]

Well in my experience, it's just highpass filtering of the AM'ed result. If you want to chase the devil right up to partial matching, you're going to need to use a synth whose base waveforms have exactly the right balance of harmonics. Might as well just get Diva or Re-Dominator in that case.

[aciddose]

Since the waveform generator in the alpha juno is entirely digital, it is a joke to think you'd need to do anything special to match the waveforms...

[itoa]

Sendy: this is not, the HPF and Juno curves look and sound very different.
It looks like the before flipping the process of filtering stops and restarts after next flipping.
btw Anybody knows how the base saw is generated?

[Sendy]

Ok, I've got it. You're right, it isn't a low cut filter, it's low boost giving the base saw it's "belly"

The alpha juno's HPF has a "boost" setting which is like a low shelf that tracks the fundamental somewhat. If you apply that shelf to the combed/am'ed combination wave, you'll get the waveshape you seek.

Without the filter boost, the waveshape produced by the Alpha is linear.

[aciddose]

I've tested the juno circuit extensively with a professional oscilloscope and modeled the waveforms. The graphs presented in this thread are a laughingstock.

The theories... I have no idea how people come up with these fantasy theories.