Synth with LFO Crossfading

[tulamide]

@Spogg
Technically, it starts with the mono section alone, glide switched off. Then a two-handed play starts, where the left hand holds chords and the right hand plays the melody. This time both, mono and poly section, and glide (which only affects the mono section) are switched on. So you hear the poly section playing chords and the melody without glide, and on top of that the mono section playing the melody with glide effects.

[Spogg]

Ahhh! Thanks for the analysis. I hadn’t thought about the glide being switched on during the demo.

Do you agree that it’s a really nice idea and feature?

Cheers

Spogg

[tulamide]

At least it's an interesting one. I haven't seen it in a VST yet. And for Kevin's Crumar clone to sound even closer to the original, it needs this "feature" anyway.

[Spogg]

Yes, I had a look at it. The Midi Mono prim and the Ruby Monostable (I hear this quite often, even in Minecraft, when they do Redstone aka "electronics", but I have no clue what it means. It's just a true false switch, isn't it?) combined is what I have in mind to do in Ruby. I just still don't get, why the env generator is in green and not in dsp? That would be way tighter regarding timing.

As you know, many of us older guys tend to use nomenclature borrowed from electronics. The names for describing the functionality are familiar to us, even though the method in programming is completely different.

A monostable is a timed switch. When triggered, the output will change state for a defined period. There are variations. They can be re-triggerable in which case the time will be extended from the most recent trigger. They could be resettable so a reset trigger will prematurely end the timed state.

I made that Env control in green just to demonstrate the requirements and to make sure the envelope generator would work as we want. DSP would have taken me longer and maybe wouldn’t have been so clear. The whole thing could also be done in Ruby of course, since sample-precision isn’t needed for an envelope generator. In fact the ADSR code is hopped at 16 samples.

Cheers

Spogg

[tulamide]

A monostable is a timed switch. When triggered, the output will change state for a defined period. There are variations. They can be re-triggerable in which case the time will be extended from the most recent trigger. They could be resettable so a reset trigger will prematurely end the timed state.

That is an eye-opener! I immediately understand hundreds of redstone contraptions now, especially those, where you get a defined amount of items by the press of a plate or putting an item in a chest (both equivalents to sending a trigger). Since items are given through something called a "dropper", which drops one item at a time (the time measure is in ticks), the initial trigger has to be automatically repeated for a specified amount of ticks, which can be easily done with a monostable that keeps the circuit powered, while a part of the circuit triggers based on a redstone clock circuit design.

That's clever, and they use terms like that all the time, so don't hold back with other terms and explanations (other than the various gates, as I use them in programming all the same). I also hear "T-flipflop" often, like "here we use a one-tick monostable with a t-flipflop". Or "pulse extender". Other terms seem to be redstone specific, like "bud switch", because bud stand for block update detector. Imagine my confusion, when I always heard it as "butt switch"

I made that Env control in green just to demonstrate the requirements and to make sure the envelope generator would work as we want. DSP would have taken me longer and maybe wouldn’t have been so clear. The whole thing could also be done in Ruby of course, since sample-precision isn’t needed for an envelope generator. In fact the ADSR code is hopped at 16 samples.

Oh, that is both, promising and disappointing. Here's why: I think there's a way to directly send Ruby Value (the red V) to blue. This will then enter the dsp/asm as a series of samples. This way I could send the first two states right after the other, with no further delay. Sample precise. But when hopping is involved, it won't recognize the second sample, right? Too bad, because I noticed that the current circuit leads to an at least 20 ms long sound output, even when ADSR is completely at 0. I thought I could reduce it to how it should behave, with some clever programming. Back to the drawing board, I'm afraid.

[k brown]

Just thought of a butt-simple if rather clumsy way of musically experimenting with playing this synth (or any) with osc 1and 2 mono; osc 3 poly. Open two instances, with one set to mono and osc3 turned down; the other set to poly with osc 1 and 2 turned down. This has the added advantage of being able to have the mono 1/2 sound and the poly 3 sound using different VCF settings. Can't believe I never thought trying this before, with any soft synth.

Edit : Just tried it and realized for this to fully work there needs to be a switch to turn off pitch bend to osc 3, so I did just that in the '_os' version and updated it's original post. Very cool to be able play a five note chord and have pitch bend (and portamento) only affect the top note. , . . and to have that top note be a completely different timbre with it's own modulations.

[Spogg]

@ tulamide
A T-Flip Flop is a binary switch that changes state on each trigger. Its most common use is to divide a frequency by two.

A pulse extender is another name for a non-retriggerable monostable. This is actually a “regular” monostable! Once a trigger is received the state changes, and during the set time any more triggers are ignored. It will only retrigger after the time period has elapsed and the initial state has been restored.

Now to the mono synth…

I was aware of the 20mS restriction on my quick schematic (10mS also works!). As you rightly said, DSP would be faster and more accurate.

The Hop(16) will actually not be an issue. Once the Env control signal gives out a 1, the ADSR starts its counting as long as the value of 3 is then seen. The ADSR counts every 16 samples and the step size is determined by the A,D and R values. If, at any time, the env control goes to 4 the release stage is started from the current envelope level. The point at which 4 is received will be subject to the hop(16) but it doesn’t matter, because the timing error will be a maximum of 360 uS.

@ Kevin
Interesting idea! But the monophonic instance won’t be high-note priority, so it’s not strictly what we’re aiming for, or is it?

Cheers

Spogg

[k brown]

Oh - right you are, I must have been hearing things.

[Spogg]

@tulamide

Here is my blue synth with DSP Env control generation.

The start value (1) now lasts for just 1 sample. Seems to work fine and just needs a valid gate ON/OFF input.

Cheers

Spogg

[Spogg]

Just for fun I’ve been fooling around with DSP and that MIDI Mono prim to give a high-note priority, with the option of re-triggering.

This won’t be a patch on what tulamide is capable of, but I just fancied having a go.

Cheers

Spogg