DSP Robotics Support

Thanks for those manuals, Bob. I knew one of them, but not the others. Man, that thing was an engineering masterpiece, an ingenious use of available resources. Comparable to the moon landing project.

Thanks, Bob and Spogg, valuable information here. But I think I have a basic misunderstanding of the concept "phase modulation".

Everything said makes sense, but as soon as it comes to the DX7-diagram of the operator, I'm still lost. Frequency data here, amplitude data there, and both are somehow combined to a new pitch value to feed the osc with. Argh.

It gets even weirder when I follow the maths-link (thanks, Spogg). Because that makes totally sense to me and is exactly what I am thinking of as phase modulation -> the maths expect two frequencies!

So, I think I'll give up on this (not on phase modulation, but on trying to recreate a DX7 operator). At least until someone shows how the operator's amplitude output is combinable with its pitch input.

Unfortunately the Yamaha DX7 diagram is pure Scheiße and, I think, is meant to confuse and mislead

I'm currently spending my 'stoner time trying to get the f**king stock multi-envelope to re-trigger so I can finish my Quilcom Blender project. If you can do that for me I'll make an operator-type module thingy to play around with.

Deal?



Spogg

Spogg wrote:pure Scheiße

You made my day!

Spogg wrote:I'm currently spending my 'stoner time trying to get the f**king stock multi-envelope to re-trigger so I can finish my Quilcom Blender project.

What exactly do you mean by not retriggering? It continues the envelope of a former key when a new key is pressed?

Hey tulamide!

Yes that's correct. Although this is not part of the DX7 thread the env may be adapted possibly to use inside an Operator module.

However, I've posted a full response in the relevant thread below to keep this one more on-topic. It has a demo synth which illustrates the fault.

viewtopic.php?f=2&t=3372&p=19876#p19876

HELP ME !!!

Guys

Here is my take on a DX7-inspired synth; The Quilcom QX7 (dev stage 38). This is orange of course

I made it for tulamide and anyone else who maybe would like to explore the sound and voice creation techniqes of FM synthesis. It is NOT an emulation and doesn't aim to replicate the old man in any way.

Here are some differences:

There is no global pitch envelope
The Operator's key scaling for amplitude is less sophisticated but from experience is sufficient.
There is no Global LFO delay
The envelope rate scaling is less sophisticated but you can make the decay and release stages faster for higher notes.
The 6 Operators are wired for Algorithm 4 only, but I've made it as easy as I can to rewire for any configuration.
The envelope generator is less sophisticated - but this makes it easier to program I think.

Some things are actually better:

Smoothly variable parameters rather than a small range of numbers in the original
16 bit rather than 12 bit, so no background artefacts (the original was terrible on pure sine waves).
The envelope decay and release slopes are exponential (thanks to Martin V!) which leads to better simulations of instruments.
You aren't limited to just sine waves so this extends the range of possibilities (not explored in any presets yet).

I've only made a few presets so far, just to give some idea.
While I was playing with the QX7 I was reminded just how powerful the technique is and just changing a couple of parameters can create quite different instrument sounds.

I hope you have fun

Cheers

Spogg

Spogg wrote:I made it for tulamide and anyone else who maybe would like to explore the sound and voice creation techniqes of FM synthesis. It is NOT an emulation and doesn't aim to replicate the old man in any way.

Thanks a lot, Spogg! I thought we were talking of phase modulation, but if even the big players call it FM synthesis, why shouldn't we, too?
From a first inspection of the schematic I can say two things: It is pretty much as I would expect phase modulation to work - and I know now what you mean with "pure Scheiße". If this is how the inputs and outputs of the operators are connected in the original, it indeed has nothing to do with the tech note. Maybe they just wanted to irritate their competitors?
I can think of quite some improvements that would make it very versatile. But I'm not sure if I'm able to realize them.

I like the rich sound that you can already get from that synth. Although there weren't 38 presets

Hey tulamide thanks for your feedback. Always appreciated.

Here's the thing: If you alter a parameter dynamically and the frequency actually varies couldn't this be called Frequency Modulation?
This is semantics (and pedantry?) I do realise, but one could argue that Phase Modulation is actually a type of FM, in the same way that modulating the base frequency value is also Frequency Modulation. One could say that this is FM by PM maybe

Regarding improvements: my QX7 is a prototype for demo and testing (like an early alpha stage) and could be enhanced and extended. The obvious one is some sort of routing matrix to produce the 32 original algorithms. Or even make the patching of the Operators fully configurable and preset-based possibly. Plus, the delayed LFO may be a valid update.
Another thing could be to provide each OP envelope with a delayed trigger; this was the main use in reality for the original's scheme. Having an attack profile made little difference musically as I recall.

Finally, here is the original paper by Chowning that Yamaha referenced in their manual and used as a source (so they say). A little light reading?

Cheers

Spogg

Hi all FM by PM fans

Here's the latest version of my QX7.

Changes are

-Master LFO now has delay and it's polyphonic unlike the original DX7. This means that every note played will individually have the LFO applied with the set delay. Nice for staggered chords and legato playing, especially for orchestral strings.

-Each Operator now has a delay setting for the ADSR. I've made 1 new preset to demo the OPs' delays and the LFO's delay. It's called Delayla (ho-ho) so see what you think... The envelope is now as flexible as the DX7's from my memory, in terms of musical usefulness at least.

-The fixed frequency controls had to be taken out of the Operators and made separate from the OP modules due to the weirdest of FS display bugs (I'll open a new topic for that one...).

I'm open to any ideas for the QX7 but I'm of the opinion the Operators are complete. I'm going to explore routing of the OPs to create different Algorithms but I think that will take me some time.

So, why not have a play...

Cheers

Spogg

Sysex to FS text file for presets

Hi

I've finished (I think) my interpretation of a DX7 in Flowstone and I decided I'd put it in its own Topic: Quilcom QX7...
If someone finds a problem I'll try to fix it but I don't intend to make any more functional changes, so the scheme is now set.

The final thing that might enhance the project would be conversion of DX7 .syx files. This wouldn't need to be in the QX7 itself, just a way of approximating original patches (presets) and changing into the Flowstone text preset format to suit the QX7 so they can be loaded and tweaked to perfection.

I am completely incapable of even starting such a project myself but, since sound design is such a time consuming process, I think it would really help to get a lot more out of the QX7.
Some areas would involve quite a translation task, e.g. converting the DX7 envelope parameters and frequency ratio to QX7 parameters. This is especially tricky since many of the original's preset values were coded exponentially.

This link shows details of the DX7's .syx formats:

http://www.yamahaforums.co.uk/forum/vie ... mat#p40019

I have many .syx files available and if some clever person would like these I can upload them, or they are easily found via Google.

So... any takers?

Cheers

Spogg