6502.org

I am back up in the frozen North for a few weeks, so once again the small breadboard and a cheap portable TV have made the bumpy ride on a somewhat questionable aircraft!

At 4:30, the school I am doing IT work at is closed so I gave the rest of the night to work on synth tech. Since I often sleep mostly only every second night, I am getting a lot done. There is nothing out here to look at besides a 6 foot snow drift, so I am highly focused.
Recently I decided to try altering the video generator to push out 256 shades of grey instead of the single bit from a shift register.
All that was needed is a bit more SRAM and to push 8 bits to an R2R DAC instead of the 74HC166.
I must say that the quality of the image is better than I would have thought.

At 640x224 and 256 shades of grey, you can even display photos....
The goal of this change is to produce a more user friendly interface with icons and to allow editing waves.
The old one bit display was good for text and lines but not much else. It felt too cold, like an IBM terminal!

I also changed the font from 8x8 to 16x16 so that text is much larger on the screen.
This isn't word processing, so a 40x14 display will be easy to read while cranking out some music.
The new font looks great at a distance, even on this 20 year old LCD gaming TV thing...
The hi res 256 shades will be great for displaying wave patterns and other cool features.
There is also a dedicated 6502 in the graphics system, so the main OS won't get bogged down with drawing graphics.
Right now for testing, I am just preloading the video memory directly from the basic 6502 OS circuit I made.

Back to work, time to get the user interface encoders up and running.

Brad

Black and white photos, keeping it real since 1843

Neil

edit: my pico video/keyboard interface chip is designed so I can use links to provide either one bit per channel RGB (so eight colours) or eight level greyscale 640x480 VGA

Yeah, it's great what you can do when you have a full 8 bits coming out... RR-GGG-BBB or my original favorite RR-GG-BB-II, or shades of grey like I am doing now.
The CRT in my synth is the same as the one in a PET or original Apple display, just a mono 12 inch tube so color is off the table.
I could have gutted a 1701 monitor for it, but then I am wandering into later 80's era parts and I want to keep it on track.

Here is basically the same output circuit (all 74HC parts) hammering out 16,384 colors on NTSC!

https://www.youtube.com/watch?v=OxsC8UrLEuM

https://www.youtube.com/watch?v=irBI9fVE9dY

On that test, I doubled up the output RAM so I had 16 bits to work with.
I sent 8 bits to an R2R DAC for 256 shades of gray and then 6 bits to the phase delay circuit for 64 NTSC color.
The result is 256 shades of 64 colors = 32,768 colors. It was just a fun experiment and it worked perfectly.
The goal was to see if I could do better than Amiga HAM mode using parts of the same era.
The end result s that I did 4 times as many colors and had no bending artifacts.

This stuff is so much fun!

Do you have updated links to your projects?
I have not seen them for some time but do remember you doing some great homebrew video.

Cheers,
Brad

Nothing recent in any of the video circuits; I'm concentrating on sorting out Tiny Basic first, then the SBC with the Pico as the video generator.

The huge snag with 65c02 and pixel addressable video is that even PAL/NTSC sizes require huge amounts of the memory space; this VGA uses 150kB... so really you need an external video processor of some sort. And I do want to use a minimum of VGA (Pico 2 could manage SVGA at two pixels per byte I think)...

What I have discovered is that in my LCD monitors, they don't give a damn about accurate sync frequencies _as long_ as the field/line ratios and sync polarity are correct.

I was working with broadcast video back in the early eighties: synchronising a 6MHz dot clock to video, to get a 32 character per line output, was easy - with half a k of text memory. So overlaying live video wasn't difficult. But broadcast spec images (PAL) were 575 lines of 52us@12 dots/us (6MHz max frequency) = 385,800 pixels.

For monochrome overlay, black or white, that's 44,850 bytes - a shade under 44k, so manageable by a 6502 if you don't want any fancy animations, in main memory. But 8-bit (each) RGB? You're looking at over a meg... you would probably sneak it to under a meg by blanking a couple of lines at the start or end of the field... but you can't keep it all in main memory.

So you have to decide which compromises you are willing to accept: low resolution blocky graphics or 'high' resolution; low colour or high colour. Or hand off the video to a co-processor of some description.

Neil

The Pico is a good choice for sure. Another contender is the Tiny4 series as it will run up to 600Mhz (cool), and offers 1MB of ultra fast RAM internally. When I am doing embedded work now I don't touch any other processor. It has the power of a PI, but lets you taste that good ole metal!

All of my 6502 video systems never map directly to the video memory. My most basic system (this one) has dual 265k video buffers which can be swapped on the vblank between 6502 access and the logic based syncgen. To write to video memory, the 6502 selects one of three 74HC574 latches - one for X Address, one for Y Address and one for Data. So to write a single pixel to the screen at 400,150 for instance, it is three writes to the hardware. Typically though, I write blocks of data so it isn't all that bad.

Here is a view of the dual memory system. Everything bellow is just 6502 IO decode using some 138 and 688 logic.
This is exactly what is shown on the photo of the breadboard above (displaying the retro DeLorean pic).
The Syncgen is a bunch of counters and 688 comparators (left input to RAMs).
It is shown as a block for now as I have the details on another schematic.

On my more complex systems that need speed (game systems shown way back in this long thread), I also added a crazy fast blitter made of 74 logic that can move blocks of data at up to 10 MHz between a third graphics ram and one of the back buffers. I don't need that on this system though as the GUI will not refresh faster than about 10Hz and is not always actively changing.

So I basically avoid compromises by greatly increasing the chip count, seems to be my thing.

Ok, here is a very basic test that shows the working rotary encoder and cursor pad that will be mounted on the main dashboard above the piano for easy access to the user interface.

The cursor buttons (up,down,left,right,fire) will be on the left and the encoder knob on the right. The cursor buttons select things and the encoder enters values.
For the friendly OS I am imagining, this will work even better than a mouse, which wasn't really such a thing back in the 80's anyhow.

Sorry for the random mess of wires and shaky video, I was trying to use the encoder while holding the phone.
This section of the board is the video encoder as shown above. I ran out of space, so the 6502 board and other logic is under the table on 2 other small boards.
It's a real bodge job, but I only took what could fit into a small tote to bring up North on my tech trip.

https://www.youtube.com/watch?v=kdn1utOAIqs

The icons are just random and I am only using 2 shades of grey in this test, but it does show how slick it will be to navigate.
The encoder is nice too, as I can cursor over to a setting like "Master Volume" and then roll the encoder to set the analog value.

I now have to give a lot of thought to the full user experience. The OS is going to be very complex, taking up multiple 64K pages.
Because I am using dual Commodore drives from an 8250, I can hold 16 full 64K pages of code on a single disk!
I will probably write 1 million lines of 6502 assembly before this thing is done too.

I guess the new name for this project is no secret anymore.... The TriAudity Computer Music Studio.
Tri because there are three working RAMs for the infinity mixer design and Audity becasue this thing is a real audio oddity.


I should probably start an official new thread now since everything up to here has just been me thinking out loud for the last 10 years.

Sounds great - please do start a new thread! One thread per project, or per major development, makes it much easier to find things. It's very easy to link across to another thread if that helps people navigate.

Tonight I found a problem that may slow development of the synth to a crawl.
.... the damn thing is too fun to play and I am not getting any coding done!

With only one cut down channel working and some temporary wiring, I can't stop jamming on this beast.
That sound... just incredible blasting out of just one of those 600 watt speakers.
My nearest neighbor is over 2 miles away, I wonder what they think of the show tonight?
Working on the part of the OS that lets me assign various sounds to the keyboard.
Everything is going to be graphical, classy, and easy to use thanks to the cursor pad and rotary encoder.
For now, the encoder and cursor buttons are just jammed in the little box sitting on top the temporary monitor.
I also had to strap another VIC keyboard to the frame since the control board is sitting on top right now.
I can load samples into the "Player" RAM right now and assign frequencies and other parameters to all 88 keys.
There is no mixer or recording yet and I have not even started on all the analog effects.
Not sure that those monster breadboards will be large enough, so I may have to build in sections then hardwire as I go along.

The 1 megabyte Micropolis floppy drives are great (Commodore 8250).
Will probably use the actual IEEE interface Commodore built as well. Good thing I left room in the center for that PCB.
The old NMOS board is damn power hungry though, so maybe my basic direct serial stream system is better. Not sure yet.

Ok, back to work (I mean jamming). Cranking out Final Countdown on the new lead synth samples I just made.
I wonder what Bill Mensch and the WDC crew thinks about all the cool projects that are still being made from the mighty 6502?

Radical Brad

Well I finally listed some of my Commodore gear for sale, starting with my Super PET Twins.
I have no room at all for anything but the synth now, so I have no choice.

Before I put this out in the wild, I am looking for some advice.
Some have told me I am nuts for not asking for a higher price, but it seems about right.
A working SP9000 in this shape usually lists for $3k on ebay, and one with minor issues not much less.

The second one booted fine 2 years ago, but now only gives the PET start sound once and awhile.
I am going to leave restoration and repairs to the lucky new owner as I don't have time. Probably a ROM or loose IC.

https://lucidscience.com/temp/4sale/spet.htm

Brad

Thanks, seems about right so I posted with a price of $3500 CAD ($2460 USD) to take both.
I know they will go to a good home that way. Being so damn rare, I know they will get the attention they deserve.
The only issue is location. No way I am shipping these, and the nearest town is Thunder Bay.
I would drive the hour to meet someone though.

Probably right on the loose connector, but I just don't have the desk room or time to even try.

On another note... added foot pedal control to the synth tonight and jammed again for 3 hours.
It is fully assignable to any parameter, but typically will act as a sustain pedal like a regular piano.

I have to force myself to finish the basic sound loader section of the OS soon so I can do a proper video demo of this beast in action.
Seems it takes more than a weekend to roll an entire graphical OS that is both useable and intuitive.
Kudos to the Amiga crew, I mean Atari crew, I mean MAC crew, I mean Xerox Parc crew... wow almost forgot who really started it all!

Brad

Heh. Took me more than a weekend to roll a working FOR/NEXT routine.

Neil

Might take that long if you had to UNROLL it and only had a text editor!

I've just been instructed by SWMBO to bake a cake, so apparently today I am $41.

Neil

p.s. not _how_ to bake, I already know that.