FXpansion releases Cypher2

[pdxindy]

1.5.3 - Pop ups, Transmod rings etc., click selected transmod slot to deselect

One of the functions of these is to give you a bigger click target area for the knob (without the knobs feeling uncomfortably crammed together on the GUI) - if you don't want to accidentally add modulation, firstly watch which ring is highlighted, or secondly, click the currently selected mod-slot to de-select, and then you'll be out of TransMod modulation mode entirely.

- Angus.

I think it would be helpful if there was a colour to the modulation ring. So it stands out more. Now it is neutral tone and just gets a bit lighter. Not so easy to see. Plus the popup still shows the knob value and not the mod value. Seems like the popup should show the mod value(s) when hovering over the mod ring. That would be a visual cue and also provide valuable information (which is currently not readily available).

In your post you mentioned that you have used Cypher as much as anyone. I'm sure if I were using Cypher for hours everyday, that I would grow comfortable with the quirks of the GUI. However, I think that a GUI should also be easy for people who take breaks from using it. Cypher could use more visual guides. For example, the transmod slots have 3 right click zones. But there is no indication at all, that this is the case. 2 of the 3 pull up the exact same menu. It is not at all obvious that one is source and the other via.

So sometimes I am traveling, or working on other projects and maybe I don't use a synth for a month or more. It is easy to start forgetting stuff that is not intuitively obvious.

1.5.5 - Removing a via source

"Vias" are really just multiplication (originally they were simply called "x", but that was too mathematical for some). So to remove a Via, choose "1" from the "Constants" mod sources.

That is not at all intuitive. So I right clicked on the transmod slot. Because there is no visual cue that there are multiple right click zones, I accidentally added a via instead of a source. Then I tried to get rid of the via. So I selected the Via menu and used Unassign option (no none option). But even though it is the Via menu, it Unassigns both. Unexpected... Okay... now I don't remember exactly which source was selected there. That source is modulating stuff still, even though it is no longer assigned. How do I find out what the source is?

So yeah, maybe for the power user, they gets used to this odd stuff, and maybe lose sight that it is kinda quirky.

Anyway... appreciate your presence in the thread. This is meant as positive criticism in the spirit of improving a gorgeous sounding synth.

[cron]

There seems to be some odd behaviour with tuning stability going on in the FM engine. Check out the sound example below. It's the init patch with only the following modifications: 'FM from 2' is turned up in osc 1, osc 2 is active but with the volume turned down (thus using purely as an FM source), sine wave mode is active on both oscs and their shape dials are turned fully left. We hear 6 notes played with very short gaps between notes, then 6 notes with slightly longer gaps between, then 6 final notes with short gaps again.

https://soundcloud.com/chqtestsubjects/ ... ning-issue

There are obvious FM tuning stability issues in the 'short gap' notes. The FM tuning 'grinds' for a moment or two then eventually settles into stable, correct tuning and stays there. As we can hear from the middle section, the tuning is stable and correct right from the beginning of the note so long as there's a big enough gap of silence before you trigger it. This example was recorded with osc phase reset switched off on both oscillators. If you switch osc phase reset on in both oscillators, the notes always grind before settling, regardless of how long a gap you leave before triggering. This example was rendered with 8x oversampling, but the same thing happens at all oversampling levels.

Please note that I'm bridging to a 32 bit DAW via jBridge, though AFAIK this shouldn't affect the audio engine. I appreciate FXpansion can't provide technical support when bridging (nor should they be expected to), so I'd be grateful if anyone else is able to confirm this happens in a 64 bit native DAW.

[Angus_FX]

I think it would be helpful if there was a colour to the modulation ring. So it stands out more. Now it is neutral tone and just gets a bit lighter.

Not a bad idea at all, I can try that.

Plus the popup still shows the knob value and not the mod value.

It does..? Here, it only does that if the mod depth is zero. If there's any modulation, it will say e.g.

(0st .. +12st)

to indicate a modulation range.

In your post you mentioned that you have used Cypher as much as anyone. I'm sure if I were using Cypher for hours everyday, that I would grow comfortable with the quirks of the GUI. However, I think that a GUI should also be easy for people who take breaks from using it. Cypher could use more visual guides. For example, the transmod slots have 3 right click zones. But there is no indication at all, that this is the case. 2 of the 3 pull up the exact same menu. It is not at all obvious that one is source and the other via.

While that's true, the right click zones aren't really there for general use - feedback on Strobe2 was that those menus (left clicks in Strobe2's case) were somewhere between unintuitive and incomprehensible depending on who you ask. So for Cypher2, the recommended way to use it is to double-click a TransMod slot (at which point the info screen turns orange & is now editable) and then use the screen area to assign mod sources: the names are much clearer in there, plus you get the visualiser which will show what the mod source is actually doing. The right-click shortcut is something I left in for power users.

That is not at all intuitive. So I right clicked on the transmod slot. Because there is no visual cue that there are multiple right click zones, I accidentally added a via instead of a source. Then I tried to get rid of the via. So I selected the Via menu and used Unassign option (no none option). But even though it is the Via menu, it Unassigns both. Unexpected... Okay... now I don't remember exactly which source was selected there. That source is modulating stuff still, even though it is no longer assigned. How do I find out what the source is?

Yes, you're right, that could be made clearer: explicit "Unassign Source", "Unassign Via" options would make sense there.

Best,
Angus.

[Angus_FX]

There seems to be some odd behaviour with tuning stability going on in the FM engine. Check out the sound example below. It's the init patch with only the following modifications: 'FM from 2' is turned up in osc 1, osc 2 is active but with the volume turned down (thus using purely as an FM source), sine wave mode is active on both oscs and their shape dials are turned fully left. We hear 6 notes played with very short gaps between notes, then 6 notes with slightly longer gaps between, then 6 final notes with short gaps again.

You're absolutely correct, we've known about this for quite some time. It's an interesting story and worthy of explanation.

First of all, it's important to consider that Cypher2 is based around an analogue model Frequency Modulation oscillator, not a digital Phase Modulation oscillator which is commonly referred to as "FM" (the classic Chowning / Yamaha DX way of doing things).

Why does this matter? For a few reasons - centre frequencies, beating, DC offsets.

Let's say you have two oscillators that are perfectly in tune, one FMing the other. No beating, just a nice clean tone.

Now detune one of them a bit, and you'll get the kind of grinding, beating etc. you'd expect.

But, equally, if the modulator oscillator has any DC offset at all (that is, its average value isn't exactly zero), it will cause the average frequency of the carrier oscillator to shift.. and then they're not perfectly in tune with one another anymore, hence beating.

Now take a variable-waveshape oscillator with phase reset, as featured in Cypher. Most of its waveforms have some inherent DC component (think about a 0-5v pulse wave on 5% duty cycle.. huge amount of undesirable DC), and there's a DC-blocking filter (a basic high-pass) to remove it. This is absolutely normal, there's a bit of art and science in choosing a filter to remove undesirable DC without removing desirable bass, but that's nothing new or clever.

But what happens when you take one of these VCO style oscillators and use it as an FM source? Well, the DC filter takes a few waveform cycles to warm up (for example, a 10Hz filter needs tens to hundreds of milliseconds). During that time, there's some DC present, which causes the modulated waveform's center frequency to shift, hence the detuning you're hearing.

In actual fact, Cypher2 (unlike its predecessor) has separate DC filters on the VCOs' audio outputs and FM paths. On the FM side, it uses a much more aggressive filter to try and mitigate the issue you've outlined above, whereas on the audio path it's set lower to preserve the bass, but even so it's not possible to remove this effect completely. In addition, when a phase reset happens, the DC filter is initialised to an approximate value based on the last known frequency and wave shape to try and get it to stabilise quicker.

One thing you might want to consider is to use one of the Ramps, set on a very short attack time, a few ms, to introduce FM modulation after the DC filter has had a chance to stabilise.

Am open to alternative suggestions, but AFAICT this is an inherent property of the synth as designed, and something you'd expect to see on anything similar.

[cron]

Am open to alternative suggestions, but AFAICT this is an inherent property of the synth as designed, and something you'd expect to see on anything similar.

Interesting stuff Angus. I recall reading somewhere that phase modulation is typically used for 'FM' because DC offset doesn't blow it up, and that PM is more 'resistant' to signal imperfections/noise in general (but particularly at the points in the chain where these imperfections tend to happen in hardware). I expect I'll still be using (rather more basic) digital synths for the 'classic' FM sounds requiring rock-solid pitch/timbral stability, but I'll give your workaround a go later on today. I bought Cypher 2 more for 'outer-limits of FM' soundscaping and experimentation at the end of the day.

While I'm here, what sort of capabilities does the analog FM model bring to the table? What can Cypher's analog FM do that classic digital PM can't?

[Angus_FX]

Yes, that's correct.

Analogue FM is just a different beast to digital phase modulation, albeit with some overlap. For a start, you've working with saw, pulse etc. waveforms instead of / as well as sine waves; you've got audio-rate PWM, hardsync and waveshape modulation going on at the same time as the FM. So whereas with a DX synth, you typically need at least four FM operators to get something interesting (the DX-7 has six, budget classics like the DX-21 and TX-81z have four), with VCO-FM you're already using harmonically rich timbres, so you can (for example) start with an already nice and complex sawtooth and use FM to add additional tonal complexity.

So it can quickly get to wilder places that a DX just won't go.. more akin to the kind of sci-fi sounds you'd get from e.g. an ARP 2600 or a big Roland or Moog modular than an 80s digital synth. But because it's polyphonic and capable of very precise tuning (albeit with the above DC-related caveats), it's equally well suited to the kinds of Rhodes and bell tones that made FM famous.

[Grizzellda]

I think it would be helpful if there was a colour to the modulation ring. So it stands out more. Now it is neutral tone and just gets a bit lighter.

Not a bad idea at all, I can try that.

Agreed. It should just be thought of as "good design practice" to use pleasant and contrasting colour combos on a GUI, especially for different functions. Or when controls are combined closely together as in the TransMod system.

And nice clear fonts are always welcomed over here!

[aMUSEd]

@aMUSEd i will ask the relevant people tomorrow & try and get an answer

No problem, I'm not complaining Just thought people might want to know there is possibly an additional discount available under some circumstances. Thanks for that.

[Tj Shredder]

Wouldn’t it be possible to construct the variable waveshape oscillators in a way, that they do not produce much dc in the first place? Your mentioned pulse train is easy to calculate what dc to subtract, to have it on average with nothing or at least very little to filter. A little dc would add to the character without disturbing too much in terms of tuning. It does not need to be perfect, but a quick estimation should be possible for keeping the modulators in tune...
Of course the classic Yamaha type digital FM does not have that problem because all oscilators are sine waves which by design do not contain any dc...

[Angus_FX]

Not really, it's inherent in the nature of a variable waveshape that the offset from zero will vary; the filter removes all DC within a matter of milliseconds (such that, per the normal understanding of the term, it can't really be considered to exist as DC at all).

The issue is specific to FM: even a *tiny* disturbance in tuning, when FMing between two harmonically rich oscillators, causes all sorts of beating and suchlike. Add in potential control feedback paths between the oscillators (A FMing B which is WMing A) and all bets are off.

[Delta Sign]

Regarding the "analog" vs "digital" FM (Phase modulation and frequency modulation would be more accurate terms).
Angus is right, phase modulation and frequency modulation basically sound the same, as long as you are using symmetrical modulators without any DC offset etc.
However, the sound is completely different once you start using more complex waveforms. The sci-fi Arp2600 comparison is pretty spot on.

For example:

FM Sweeps

• The first sound is digital FM with sine waves
• The second sound is the same digital FM sweep with an asymmetrical modulator
• The third one analog FM with sine waves
• The last one is analog FM with an asymmetrical modulator

You can really hear how the sound starts to "fall apart" on the last one. That may be a thing you want, or not.

[Tj Shredder]

Not really, it's inherent in the nature of a variable waveshape that the offset from zero will vary

Yes, but with the example of the pulse train, you know it mathematically pretty precise, it is directly correlated to the pulse size in %. Other waveforms might be more difficult to calculate, but you could even create tables prepared to correct the to be expected dc...

the filter removes all DC within a matter of milliseconds (such that, per the normal understanding of the term, it can't really be considered to exist as DC at all).

If the filter should pass 20 hz unaltered, its about 50 ms, which is a lot. If the oscillatores are near zero dc, that doesn’t hurt as the detuning is tiny if at all.
If you adapt the filter to the lowest osc frequency, you’ll be much faster, and on really low notes the tuning in doesn’t hurt so much as well (bass players are always out of tune as you probably know and nobody cares too much...; - )

[Angus_FX]

The point is that tiny detuning, which would be inaudible on its own, is harmful when dealing with harmonic-rich FM.

A better way could be to initialise the DC filter (on the FM path only) at a higher cutoff frequency and quickly sweep it down. Further investigation required.

[Stefken]

A skin I'm working on.
It would certainly be nice to have some explanation on how to skin Cypher 2.

Is there a way to target specific knobs? I wanted to draw attention to the filter cutoff.
If I change the colours of styModKnobLarge3, I can change the color of the knob, but others change as well.
I'd prefer more granular control.

Is there a faster way than deleting the plugin and reopening it in your DAW, to see the changes in the xml reflected in the GUI ( I use Ableton 10 on Windows 7, 64 bit)?

[roland_rock]

Right, two weeks and 534 forum posts later, I'm back. Going to try and answer as much as possible in one big post, apologies if I miss anything.
...

1.3 - Windows 7

Honestly I wasn't aware that it's still such a popular OS. The detailed story on this is.. detailed, read on.

Strictly speaking, the minimum API level we support is Windows 7 with SP1, and the post-SP1 "Platform Update" installed. This is, mostly, because Cypher2 uses DirectWrite and Direct2D for its graphics renderer, which wasn't available before that update. (Use of these APIs is necessary for hardware accelerated vector graphics, and High DPI monitor support).

However, because we can't guarantee that every Win7 system will have this service pack and update installed, we can't simply say "supports Windows 7" - it won't run on a vanilla Win7 install, for the reasons outlined above. OTOH, we had a lot of feedback about this one, so I'll look in to whether the messaging there can be changed.
...
Thanks for reading - feedback welcomed.

- Angus.

Hi Angus, hope you had a nice break
So... is Windows 8.1 supported for Cypher2? I'm still running that, and moving to Windows 10 will be a big pain (many dozens of plugins to re-authorize), so I won't be doing it for a while. I would love to upgrade to Cypher2 (from Dcam Synth Squad), but don't really want to fork out if it will effectively be unsupported (says Win10 min on web site).
Ta!