Colour palette and sound.

[rampa069]

Hello Istvan.

Tank you very much for your fpgated project.

I have ported it to others FPGA and i have made some changes that may be of your interest.

1) The color palette was a bit wrong compared to the real machine. I have translated the yape (emulator) palette. As i have FPGAs with 8bit colours, the lut translation is 24 bits. but it is also ok on less bits. If you want to use it, you have the changes here

2) On sound i have made similar changes to output to a 16bit dac. Eliminating the need of a filter on the output. Also logarithmic volume. changes here.

Now it looks and sound very close to the real machine.

All the software i am testing performs OK. only a few glitches on some demos. This one as an example. But not sure if the problem is in ted or in other place (may be IRQ).

Thanks again.

[ishe]

Hi Ramon,

I know that the RGB palette that I have used is not perfect. Honestly I had no time to investigate that part so I just took these values from a friend and used that for developing fpgated. It was my intent to investigate it later using colodore's sciense. Last year I have measured color phases via oscilloscope and I have used those values to create a CVBS signal generator to have a drop-in TED replacement. That works quite well. Recently I have done the DVI implementation so that I can now hook up a digital display too. That uses the old color values so I will investigate your colors from Yape and try. Have you checked colodore against yape colors?

Regarding sound I wanted to have exactly the same signal out of the snd pin as the original ted, thus the PWM. As I am now working on sending the sound via HDMI I might make use of your 16bit dac implementation. I need L-PCM for HDMI and the current PWM is not good for that.

Although most programs work fine there might be still some incompatibilities like the demo you mention. Last year I have eliminated two bugs so please check in your ted module's header what is the version. If it is 1.3 or higher then that has the fixes. I will test this demo too.

I have now a working prototype of the TED drop in module (with RGBS and HDMI headers) so I can test it in a real C16 or Plus4. I will add some project notes soon on hackaday with photos!

Thanks for the findings! Istvan

[rampa069]

great news!

In your new , if you are going to use HDMI audio, you will get what you need. 16bit sound and 24 bit color information.

Didnt test far more than comparing the image from the fpga and a real C16 on the same screen. and they are the same colours. Same with the sound.

If you need the 1bit PWM, always can put a simple dac inside the module and get 1bit pwm from the 16 bits. i have done this on mist and the sound is clear without the iir filter.

About the "buggy" demo.....

Was testing tonight and it is not working. Same glitches on yapesdl and vice. but works ok on yape (win) and real hardware. Curious if you can test with the dropin replacement to see if the problem is on TED or in other place.

What emulator do you use for debugging? would be great if i can get a emulator with sources that work well with this demo.

Thanks.

[rampa069]

This demo i thikn is using an IRQ loader (krill loader). In the loader there is a disk image that says"buggy emulator"... The sources of the loader (and the .d64 for testing) are here. have to investigate the source tosee what is looking for...

[ishe]

Hi, I have tested it on my fpgated prototype and a C16 PAL machine with 1541-II floppy drive. The demo run however I have seen some problems which will need further investigation. I suspect the raster irq mechanism needs some review and deep analysis. I will check krill loader. Interestingly this demo does not run in plus4emu just in yape. I can however investigate yape code also to see what can be different in raster irq. The C16 I have used has a real 8501 CPU. Unfortunately I am at the moment busy with my HDMI implementation (as I have mentioned DVI works) but after that this is the next step.

[rampa069]

have tracked down the krill warning to the c1541 implementation (it thinks its a 1541U) so. i'll continue trying to investigate...

[gyurco]

I believe this fixed Krill's loader on C64: https://github.com/mist-devel/c64/commit/b5ab74ae69eb338e40289ea36a070958a331d0d1