The Sinclair QL Forum

This is the QLion Gold edition card, final version at least for a long while.
Improvements from my previous cards are the much faster performance due to the 16bit ram access, 4MB of ram instead of 2MB and digital sample playback.


It has

- 68000 fpga core
- Fast dual bus video ram that together with the faster cpu makes QL perform in average more than 10 times faster than the standard.
- 4MB of ram, 3968K available to OS, 128K as extra video ram for the 256 color modes
- 512K flash rom lets you choose from 4 sets of rom and driver rom (JS+qlsd, Minerva+qlsd+tk2 etc)
- QLSD storage for win1_ to win8_ drives. It's driver works well with toolkit 2.36, has some issues with some versions of toolkit.
- VGA up-scaled output that can be easily converted to hdmi if needed with a cheap external adapter (QL standard video output is not usable)
- 16 colors for mode 8.
- 256 color graphics modes for both 256x256 and 512x256 resolutions that can optionally be mixed with the standard QL modes, screens of standard and 256 color modes are drawn independently and can be switched instantly.
- 46 hardware sprites of 16x16 pixels and 15 colors + transparency, great help in creating games in C and even in Basic.
- PS/2 mouse that works in the pointer environment by loading a driver
- 3 channels of audio + noise emulating the standard AY-3-8910/YM2149 ic, 3.5mm stereo audio jack output
- 8bit stereo digital sample playback at configurable sample rate, only 2K dedicated buffer because I run out of bram, small but in practice enough for the simple player I wrote.
- fast serial port with 3.3v lvttl output to use with a common serial to usb board at speeds much more than 115200, 1K transmit + 1K receive hardware controled buffer.
- battery backed up real time clock using a DS1307 ic
- Execution speed control in 64 steps for playing games or accessing microdrives.

This is not a commercial product, I may produce cards for those interested at a slow rate. It's a use at your own risk thing.
Also there is not much software for some of the new features, only some test basic programs, a galaxianish sprite demo game written in standard C, a ym file player and a digital sample player.
I also made 3 utility programs for fat/win file transfer that are very useful. One that displays the directory of the sd-card, one that copies a file from the fat system of the sd card to a file at the win drive and one that copies a file from the win image to one at the sd card.

It's in testing process but I see no problems with my issue 5 and issue 6 QLs. It doesn't support Minerva second screen.

Leon

Congratulations! I'm really stunned at which rate you design new PCBs!
Speedwise you could call it SuperGold edition, at least according to the Q-Top benchmark.
May I ask at what frequency you run the TG68K.C to get that result?

Hi, thanks, I use a clock at 32Mhz but it differs how many clock cycles of this clock take each different type of access, i.e. rom is still 8-bit.
This pcb is similar to the previous so easier to make. Also it is big enough to make routing easier.

This new version looks great. Very nice job. I am interested in buying one when you produce them.

Can you please add me to the potential users list?

Thank you
Fede

ΟΚ

lliont wrote: Tue Dec 03, 2024 11:50 pm Hi, thanks, I use a clock at 32Mhz but it differs how many clock cycles of this clock take each different type of access, i.e. rom is still 8-bit.

Sounds like a performance problem which you could fix without designing a new PCB.
I'd take a Q40/Q60 like shadow RAM approach. Which means:
By default, read access goes to ROM, but write on the same address goes to a shadow RAM area.
In an additional mode, read access goes to shadow RAM, write-protected.
This infrastructure would provide full speed for the ROM area if desired.
Not sure how much it would impact your Q-Top index, but there are certainly some OS dependencies in that benchmark.

Yes Peter I can do that, I thought a similar process, the fpga to initialy copy the rom contents to ram and then make this ram write protected and then start the system.
I have used this aproach in another core but now I have used even the last logic gate and I must find a way to squeeze this in the fpga.

On the other hand even if I do nothing more it is fast enough already.

Great work as usual lliont.
I already have too many computers here to think of buying any more systems, but I watch your work with interest.

I see an opportunity for bespoke "QL furniture"

I do like the look of this latest 4MB offering - I hope it moves towards becoming a product...