Repro-1 (out now)

[egbert101]

<delete>

[david.beholder]

U-he Repro-1 vs Pro-One

Interesting. Re-pro-1 square is horizontally straight and Pro-One square has horizontally falling. Is it because of old capacitors?

[Urs]

U-he Repro-1 vs Pro-One

Interesting. Re-pro-1 square is horizontally straight and Pro-One square has horizontally falling. Is it because of old capacitors?

Don't think it's old caps. Maybe a tad of a highpass/DC filter in mxing desk or audio device. Or maybe the highpass behind the filter is tuned sightly higher. Maybe Sequential ran out of these parts and used those instead (no kidding... just look at the first batch of Rev3 Prophet 5s which had all sorts of Rev2 parts in them, just because, well, they still had so many)

[egbert101]

<delete>

[mcbpete]

(with the spikey bits)

To an idiot (like me!) what are those spiky bits that you saw in corners of the waveforms in Repro-1 but not the Pro 1. I was actually surprised it was that way round - I would've totally expected the analogue synth to not be able to create a clean waveform like the Pro 1 manages (but oddly the Repro doesn't in those tests)

[chk071]

Isn't that the anti-aliasing filter at work?

[david.beholder]

(with the spikey bits)

To an idiot (like me!) what are those spiky bits that you saw in corners of the waveforms in Repro-1 but not the Pro 1. I was actually surprised it was that way round - I would've totally expected the analogue synth to not be able to create a clean waveform like the Pro 1 manages (but oddly the Repro doesn't in those tests)

It's Gibb's Artefacts (see: https://en.wikipedia.org/wiki/Ringing_artifacts) and i think they are more noticeable in case of U-he Re-pro-1 because:
1. Different gain level of synths. U-he's one is way more loud according to the pictures
2. Generally more low-passed content from that particular Por-one (noticable in 12k vs 18k)

Isn't that the anti-aliasing filter at work?

Doesn't look like one if both of scopes are working one the same [project] frequency.

---

Actual analog synths frequency might be unfiltered but input device (adc or pre-adc) should have filter to prevent higher than half-samplerate frequencies appearing in digital content. This filter might work a bit differently to U-he's downsampling filter with the same purpose. But GA should be presented on both spectres due to band-unlimited nature of osc signals.

[egbert101]

<delete>

[twal]

Edit: sorry, got too excited

[Urs]

(with the spikey bits)

To an idiot (like me!) what are those spiky bits that you saw in corners of the waveforms in Repro-1 but not the Pro 1. I was actually surprised it was that way round - I would've totally expected the analogue synth to not be able to create a clean waveform like the Pro 1 manages (but oddly the Repro doesn't in those tests)

A few observations:

• when you put an oscilloscope (hardware) behind a Pro-One, there's no spiky stuff at all
• when you run a Pro-One through a sound card, there'll be jaggy bits on both sides of the edge
• when you run Repro-1 at 44.1kHz or 48kHz, there'll be large jaggy bits on the right hand side of the edge
• when you run Repro-1 at 192kHz there'll be small jaggy bits on both sides of the edge

Now, the reconstruction filters in high quality audio interfaces are symmetric, i.e. "linear phase". They create those symmetric jaggy bits on both sides of the edge of a sawtooth or square wave, simply by bandlimiting the signal. So, any perfect analogue waveform will get jaggy stuff once recorded through a good audio interface. It's a sign of good sound quality.

The synthesis method in Repro-1 uses a similar, linear phase technique to bandlimit the edges of its oscillator waveforms.

When run at low samplerates, Repro-1 needs to oversample its filter more than when you run it at higher samplerates. At 44.1/48kHz it's oversampled 8x, at 88.1/96kHz it's 4x and beyond that (192 etc.) it's twice. For realtime purpose, the oversampling is latency free and thus not linear phase. Which is totally indifferent because there is no dry/wet signal to be mixed with, unlike mastering level linear phase equalizers. It's indifferent furthermore because humans can't hear phase relation of high frequency harmonics, hence there is no audible difference.

Nevertheless, visually there is. The low-latency oversampling shifts the phase beyond 6kHz or so by a fraction of a millisecond. Visually, it moves the jaggy bits from the left side onto the right side of the waveform, where they build the cluster we see. The less oversampling, the less of a cluster.

There's one issue though. The wavefolder sounds slightly different at different samplerates, due to this (as mentioned in the manual).

- Urs

[Urs]

(that said, I'll need to check if Repro-1 waveforms are really less jaggy at 192k... half a year ago they were, but who knows what we changed during optimization before release...)

[Chris-S]

There's one issue though. The wavefolder sounds slightly different at different samplerates, due to this (as mentioned in the manual).

And what about cross modulations (modulating one osc with another @audiorate)?

[mcbpete]

Thanks Urs for the awesome explanation - That does really help my understanding of it ! Does the waveshaper sounding different at different sample rates have any relation to the LFO running at different speeds at different sample rates ?

[Urs]

There's one issue though. The wavefolder sounds slightly different at different samplerates, due to this (as mentioned in the manual).

And what about cross modulations (modulating one osc with another @audiorate)?

It's the filter that does it, not the osc.

That said, the analogue oscillator is a bit jaggy as well, but at extremely high frequencies (megaHertz). That's from the Opamps swinging into an equilibrium between ports (-) and (+). This ringing does not pass the filter though, you gotta put the oscilloscope right inside the box.

While there might be a difference in sound, it's pretty much on par with the difference between two Pro-Ones. We could get FM to sound exactly like one reference, but then not like the other until we tweaked internal settings - once it sounded like the other, it didn't sound like the first one.

[Urs]

Thanks Urs for the awesome explanation - That does really help my understanding of it ! Does the waveshaper sounding different at different sample rates have any relation to the LFO running at different speeds at different sample rates ?

Does the LFO run at different speed in different sample rates? - That would be a bug then... for which we don't have a ticket in our system, hmmm...