My Hardware SID Synthesizer Project...

[Thorsten Klose]

D.I.Y.

Best Regards, Thorsten.

[DHS]

Well, i think i'll build one of those beast.
The demos are impressive enough

The only problem for now is that i have 2 JDMs, and, of course, none of them support the PIC18F452 format

Time to buy another one i guess :-/

Have you ever thinked about Atmel, scenix and othe MCP like those?
(just for curiosity, nothing more).

Another nice touch could be the addition of a little musical keyboard

Cheers.

[Thorsten Klose]

Cool!

The only problem for now is that i have 2 JDMs, and, of course, none of them support the PIC18F452 format

Time to buy another one i guess :-/

If you own a programmer based on the original plans from JDM, there is no need to buy a new one. See the troubleshooting topic of this page: http://www.ucapps.de/mbhp_jdm.html - you only have to add a 10k pull-down resistor to pin #38. And if the programming still fails: just add an external power supply.

You only need the JDM to burn the bootstrap loader into the PIC, thereafter the operating system as well as the application can be uploaded via MIDI in-circuit and within seconds (http://www.ucapps.de/mios_bootstrap.html)

Have you ever thinked about Atmel, scenix and othe MCP like those?
(just for curiosity, nothing more).

I'm using the PIC due to historical reasons, but compared to Atmel AVR it still has advantages: the PIC16F877 which I used for the old designs provides superior features like a 14-bit word flash (-> means 2 7-bit MIDI events can be stored in every cell), the possibility to program the flash from internal (-> via MIDI w/o programmer), and a superfast ADC which converts pot values appr. 6 times faster than the old AVR devices which were available at this time - this was essential for a MIDI controller.

Advantages of the PIC18F452: pin compatible with PIC16F877, so that the existing core module can be re-used, much more flash and SRAM, about 2-3 times faster, improved architecture (e.g. multiple pointers, no code banking and one-cycle multiplication), etc. etc...

Other advantages of PIC compared to other uCs: DIP package (no SMD), very reliable, hard to damage, worldwide availability for hobbyists (low prices, small quantities).

Another nice touch could be the addition of a little musical keyboard

only problem is to get the claviature :-/

Best Regards, Thorsten.

[Razmo]

Hi again Thorsten! :O)

I've got three more questions to ask you about your device before rushing to make one myself. I'm seriously considdering if it is worth spending the time creating an alternative OS for your device, but for me to take the decission not to, I need to know this:

1. Have you implemented a feature that allow one to restart the phase of the waveforms on every note? This I find to be VERY IMPORTANT, as this allows for some REALLY nice patches (I'm thinking about toggling the TEST bit on/off in keysync). If not, then PLEASE IMPLEMENT THAT! :O)

2. Can the envelopes attain an exponential curve?... I've found that the most useful basslines can be created when the DECAY especially, is exponentially modulated. I'm thinking of a variable control that allow you to set this curve from linear to very exponential (up or down). I made this on my own engine, and it's essential to me. I do it in a simple way, where I simply feedback the EGs output to modulate the EGs decay rate, and that works fine, but maybe some other methods can be used (?)

3. Is the new 8580 filter impr. you mention on your site an adjustment for the fact that the 8580 filter is controlled like the freq. registers? I know that the 8580 filter sounds awful when controlled linearly. This is a MUST!

If these three things will be, or are reality, I'm most certainly going to use your device/OS, otherwise I'll see me forced to make my own OS anyways, cause these things were the reason I started on making my own eventually.

Regards, Jess D. Skov-Nielsen.

[Thorsten Klose]

Hi Jess,

I think that you don't need to program a completely new software engine just to implement such features. Why not integrating this into the existing MBSID engine and releasing it to the public? (thats the idea behind open source and especially of the GPL). Of course, nobody prevents you from replacing the whole MBSID engine by alternative routines, or to write a completely different synth, but keep in mind how much developing effort is required to implement all the other (not less essential ) features which are already provided by the existing application.

1. Have you implemented a feature that allow one to restart the phase of the waveforms on every note? This I find to be VERY IMPORTANT, as this allows for some REALLY nice patches (I'm thinking about toggling the TEST bit on/off in keysync).

see sid_sw.inc - the test bit could be toggled within the SIDSW_MACRO_NOTE macro, behind SIDSW_Note_GateSetReq. Or better: request a test bit reset, so that the SID_SR_Handler in sid_sr.inc can do this before setting the gate bit. It possibly requires 10 lines of additional code.

2. Can the envelopes attain an exponential curve?... I've found that the most useful basslines can be created when the DECAY especially, is exponentially modulated. I'm thinking of a variable control that allow you to set this curve from linear to very exponential (up or down). I made this on my own engine, and it's essential to me. I do it in a simple way, where I simply feedback the EGs output to modulate the EGs decay rate, and that works fine, but maybe some other methods can be used (?)

Could also be included into sid_sw.inc - or not?

3. Is the new 8580 filter impr. you mention on your site an adjustment for the fact that the 8580 filter is controlled like the freq. registers? I know that the 8580 filter sounds awful when controlled linearly. This is a MUST!

no, just only the output value of the modulation matrix will be divided by 4 (*) to improve the "influence" on the 8580 filter - search for CBM8580_FILTER_FIX in sid_sw.inc
There is also a .mp3 which demonstrates the difference (mbsid_v1_5_filter.mp3)

(*) I tested it also with an logarithmic curve but noticed no audible differences.

Best Regards, Thorsten.

[Razmo]

Hi again!

I guess I had to comment on that way you handle the 8580 filter. I can see that the effect will be better this way, but what I did was to use the self same table as for the freq. notes. As the SID 8580 filter is tuned in the same way as the Freq. regs you will obtain a VERY linear response from the 8580 filter. The drawback with this technique is, that the resolution of the filter is much more dimminished in the lower filter freq. ranges. In fact You'll end up with a table that has only 128 entries for the cutoff, otherwise many of the lower values will be identical. Even though, the result are really impressive to listen to! and with a logarithmic response added to the EG controlling the 8580 filter really beef the sound up additionally. It has to be heard to understand how it sounds... I'd say that the basslines get a hell of a lot more "bouncy" with this technique.
The only drawback of the 128 step filter is showing when you use very slow attack and release times, as you can eventually hear the stepping in this case, though the rates would have to be VERY long before it gets noticeable.

Regards, Jess.

[Thorsten Klose]

Hi Jess,

in fact I used the note table for this test (it was the most simple way ) , but it didn't sound so good like the current implementation.
Did you also try it in the same way like me?

Best Regards, Thorsten.

[Razmo]

Hi!

You must have done something wrong, because it's still in my engine, and it sounds great! ... Did you remember to take the top 11 bits of the freq. table values when inserting them into the 11bit filterfreq regs?

[Razmo]

Hi!

If you've not already been there, then check out my three 8580 filter examples at http://www.kubarth.de/sid/index.html and hear how the 8580 filter sounds when controlled with a 128 value table and a logarithmic curve for the filter EG.

[Thorsten Klose]

Hi,

ah, now I remember from where I know your name...
Great tunes! Great demonstration!

Best Regards, Thorsten.

[Razmo]

Ahh! I see! ;O)

Actually I can Email you some more examples than those that I've got lyin' around here on my PC, so that you can better hear what I'm thinking of?

[Thorsten Klose]

I'm up for it!

I'm especially interested in the notes and controllers which are sent to the synth (do you still own the MIDI sequences?), and I would like to get an impression about the voice and envelope settings. I"m not a SID expert like you who can 'read this out' by hearing a tune.

And of course: it would be a pleasure to have the same algorithms in MIDIbox SID as a switchable option!

Best Regards, Thorsten.

[Razmo]

Hi again Thorsten! :O)

With what you've done, I'd definitely say you are a SID expert, at least when it comes to Hardware and coding. My force probably lies in the fact, that I'm rather experienced in programming synthesizers, and thus know what I want! ;O) ... I'm currently using:

Korg Z1
Waldorf Micro Q
Waldorf Pulse
Waldorf Microwave II
Waldorf RackAttack
Oberheim Matrix-1000
Yamaha FS1R
Yamaha VL70M
EMU E6400 ULTRA w. RFX

Programing all that gives me a lot of insight! :O)

Actually I'm extremely novice in Electronics, which is also why I'm REALLY interrested in your device, because I then know, that it'll be quality hardware, instead of my own HW which I'm a bit uncertain of. I'd REALLY like to be involved in getting those feautres to your design that I spoke of, and will do anything I can to help out on it.

Unfortunately I do not have the MIDIs for those examples, but you know what!? ... I'll make some total new ones without added FX, email them to you incl. the MIDIs. What do you say about that?

By the way, I'm also VERY much interrested in your comming OPL3 project. Mainly this is because I've once owned two old KORG synths (DS-8 and 707) which was early FM synths made simple. They had one force that I've not seen on ANY other FM synth these days (not even my FS1R!). They had the option to switch the machine into 8 voice unison, that was REALLY awesome! ... They could fire off some FM basses that simply whent KAPOW! I tell you. I'd REALLY like to see that in your project!!! :O)

But anyways, I'll mail you the rest of my MP3 examples of my current code.

[ifadeo]

hi thorsten,

here a short link about your project at the Forum-64 ....
only in german...sorry...

http://www.forum-64.de/wbb2/thread.php? ... =2371&sid=



cheers ifadeo

[Razmo]

Holy shit! :O)

Ifadeo How could you do that! ... show us such a link!?
Now I feel like a serial killer, instead of a guy who tries to give new life to a shining little SID baby, by donoring a dead C64 SID.
I'm a doctor experimenting on little mice... ;O(

Hope I'm allowed to reply on that thread in English... Got to explain myself!