6502.org

And all that dark matter in the universe? Foam packing peanuts.

Random musing...

I wonder if I could bypass the huge mainboard in the 8050 disk drive and just interface directly to the Micropolis 1006 drive?
The Commodore 8050 mainboard includes dual 6502s and is basically a fully working computer by itself.
In my system, I don't need so much complexity, just a simple stream of data. I could even do without DOS altogether.
With a catalog of samples on another disk, I would only need to know the start and end sectors on the disk for each sample.
Such a scheme may be much faster and actually hold a lot more data.

This idea comes from reading over what WOZ did with the original Apple floppy. Genius actually.

Here are the Micropolis drives. Two drives share the single controller, which does step drive and head decoding...
Here is the 6502 computer inside the drive that allows the drive to act as an independent computer basically...
There are no black magic analog components on the mainboard, just basic digital, so I think I can do the same thing with my own mainboard.
All samples will be basic streams between 100k and 512k for the most part, so I really don't need all of this DOS complexity.

I will be giving this some serious thought and might do some tests this winter.

Brad

I 100% agree with everything you just said!!

But... I will dive off a cliff before I consider using any modern tech to spoil this pure design.
No CPLD, no microcontrollers, no LSI at all.
I will also not allow any work to be done outside my extremely limited shop no matter how much time and effort it would save me.
No way, never. I will drop this entire project in the recycle bin if I can't to it all 100% by myself using age appropriate tech.
You are not speaking to a normal person here... so the only logic that works is the 7400 on my breadboard!

I should have waited to post actually.
I have already found it quite trivial to read the binary stream right form the drive and control both the steppers and spindle motor.
Seems the Micropolis board already does most of the heavy lifting. It's all digital in and out!
Right now, I have the drive acting much like the VIC-20 dataset would.

This fits my original design ideas much better actually.
I wanted to have a file on the main OS disk called "Catalog.bin".
In this file will be basically the simple index to the disk number and locations of samples.
How it works is simple and fast...

In the OS when it comes time to loads samples into a channel RAM, the OS will read Catalog.bin.
At that point, the OS will prompt... "Insert Disk Number 013 and press Enter".
The controller will then know to just stream direct from a certain sector range.
Fast and easy.

As samples are added or deleted, the OS will update the catalog.
No need for any file system on the actual disk, so more room for data.
I will do more work on this soon, but already have the disk spinning and showing binary streams.

As for the use of modern tech to make my life SO MUCH easier....
Did I mention I am not normal? Nope, NEVER going to happen.
I can't even tell you how much I am looking forward to spending 8 hour long days soldering all those sockets point to point!
Debug sessions that last days just to find one wrong wire, up until 4:00am to finish a small section, endless caffeine while listening to 70's synthwave... yes!

I live for this 83-72-73-84!

Nice!
Considering we are all here to pay tribute to a 1970's CPU, I guess "normal" would actually be an insult.

We went from green grass to 4 inches of snow over the last 2 days so I finally brought the Synth into the house.
I am far enough along to take inspiration from the frame and can clamp some scrap plywood on the front to mount the CRT.
Final design of the digital channels and the start of the analog filters section will now commence.

Getting the frame indoors was quite a challenge actually. I feel like I just did a full body workout.
My welding shack is about 2000 feet from the house down a very steep hill, now covered in snow.
The small cherry trees along the trail are not so small now, so I was banging through the bush with this beast.
Up hill, in the dark, through the bush in the cold... just like your grandpa used to get to school.

Oh, and I have a tall house, so it's 15 stairs up and then 15 stairs back to the basement.

I tried hammering out some 80's synthwave tonight, but only got silence.
Maybe I am too tired.... or maybe I need a circuit board with 1200 ICs first.

Exactly the same footprint except that this is 10 inches wider.
Looks good along with my 1890's piano stool. Will have to refinish it now.

Just like _I_ went to school... though to be fair, I rarely dragged a piano along

Getting a nice feeling for the size of the beast now.

Neil

I have had a few days to work on the most complex part of the Synth... the operating system in the 6502 that runs the tracker.
Originally, I was moving towards 16 full channels (each with its own 6502 and RAM), and did manage to get 2 basic channels working.

Then I woke at 2:00am one morning and had a crazy idea that would allow an infinite number of tracks with only 10% of the RAM I would have needed for 16 tracks.
This idea seemed too good to be true as it would also reduce the number of 6502's needed from 18 to only 3 or 4.
It would also allow for individual HQ samples to be called for every single key on the piano instead of pitch changing in blocks.

The idea : Use only one HQ channel and tracker, but render and layer to dual recording RAMs.

I realized this obvious idea when I was using a modern DAW to make a music track on an ad the other day.
I always start with a full drum track, then do some bass, then fill, then maybe lead, etc.

So instead of trying to control and sync 16 massive channels and tracks, I only need one!

It was a lot of wiring, but I strapped up a proof of concept with one channel having 16MB of SRAM, which is enough to load an entire sampled piano with great quality or a drum kit with 50+ hits.
I then wired up dual 16MB SRAMs on 32KHz counters, one that allowed recording from the output of and ADC fed from the channel and the other that allowed playback to a DAC mixed back with the live channel. I am only using 12 bit samples for now, but will probably move to a 16 bit DAC now that I only need a single ADC and DAC in the entire system.

Keeping the output from the live channel to 12 bits into the recording RAM will allow for more mixing headroom if the final output is 16 bits.
This configuration will allow even more room for the live keyboard samples and allow full volume output up to 16 layers before clipping out of a 16 bit DAC.
So even a 32 track song is completely doable by keeping the max live channel volume just below the clipping level. 64 channels? Sure! This thing is infinite.

This trial worked so well that I am now moving into the final design stage on this Infinite Tracker!
I don't know why this was never done in the 80's or 90's as it opens up an endless ability to create thick layers of music.

It also allows easy storage on my 5.25 floppy disk drives as I am only saving the track data.
Samples are loaded from disk to the piano, and tracks are loaded to the tracking system in the OS.

For full editing of merged tracks, the system does a "rewind" of time, starting by remixing up to the point you want to start over.
Yes, this takes as long as a track in time, but still... this is 1980 after all and still way better than lossy tape editing.

The end result of this new design is a full production ready song without using any external mixer or analog tape!

Here is a very basic 5 layer test I did just now. Tried for an 80's Disco+SynthWave type of thing.
Keep in mind that there is no quantizing or tracker done yet, I just riffed out live into the RAM banks with my keyboard hanging half off the desk.
I used 5 different samples loaded in mono 12 bit format so that each used piano key had it's own sample.
Some "sample packs" might have one for all 8 keys and some may only need an octave or two. It's really quite versatile and efficient this way.

https://lucidscience.com/temp/SynthTest1.mp3

This sounds so much better than my original frequency changing system.... no aliasing or timbral oddities.
The first design used a 20MHz clock fed into dual 8 bit counters and comparators to set playback frequency, so there were artifacts as you move too far from the original sample.
The Fairlight CMI had this strange effect, which is what gave the Terminator II score it's robotic sound.

Here is an example of the frequency shift design I did 10 years ago. This was only 8 bit into a cheap resistor DAC as it was only for a game system.
Dual 74590 counters fed by 16 MHz being reset by values in dual 74688 comparators. Simple, and still better than a phase accumulator for sound quality...

https://www.youtube.com/watch?v=fuJz3ERi8Kg

I also have no effects working yet, and plan to include echo, distortion, lowpass, resonance, pitch bend, etc.
Since there is only one live channel, I can really go all out on the effects chain as I only need to build one and not 16 of them now.

When the tracker is done, it will have note level editing, quantize, FX controls, etc.
This project just took a massive upturn in quality. Wish I was actually living in 1984 - it would change the industry.
I am still using ONLY tech that was available in 1980 for this project, including only DIP ICs.
So yes... this could have been done in 1980.

Oh well, better never than late.

Will post more videos and photos of this beast when it is not a huge mess hanging off my desk.
I also need to chill out and make a schematic of where I am going before I get too far ahead.

Happy Holidays Retro Freaks!
Radical Brad

I don't know which 1980 you were living in. It wasn't until the mid-90s that I could afford 16MB of RAM for my computer. With a 16Kbit DRAM at about $10US in 1980, that's 8192 chips costing over $80,000.

Still, it's a very cool project and this is a big step up in its capabilities. And while it's something few people could afford, it is technically possible for the time.

Thanks!

My competition of the era is the Fairlight, which at the time cost $100,000 for base unit. You could go with a Sinclavier for $250,000 as well!
This will cost much less than either of those and have a lot more capability including a built in recording studio.

Since I have been hired by the Admiral Computer Corporation to commission this prototype, their ability to mass produce RAM chips in-house will keep costs lower
They are also supplying the dual disk drives and CRT core from their PAL-2001 series computer and the keyboard from one of their manufacturing partners.
CEO Jacques Leimart has also promised that by the time we bring it to market (1984), RAM will have made a huge price drop. Upgrades, upgrades!

The modular nature of the design also allows populating as much RAM as budget allows from 4MB base to 16MB max.
The demo song I just did used about 2MB.

I shall address this in my video series.

Brad