Could RPi be used for Atari HDMI output? Like this:

[Simius]

On 1/13/2021 at 7:12 PM, ivop said:

Existing software that uses the E: handler will automatically use it. That would be my main goal for a cheap 80-column hardware solution. Sophia2 cost me €65,= including shipping with a group order. If I can get the same for $29 US and a Pi Zero, I'd choose that. And no need for expensive DVI to HDMI cables. Which reminds me, I still have to order such a cable

 

If you want 80-column hardware solution working with a special handler, you just need an existing, genuine Atari XEP80 device. If you want to convert a color image generated by the GTIA to the digital image, you don’t need any Pi Zero nor an adapter for $29, but an ordinary S-Video cable for a few bucks. Your LCD TV will convert the image itself. Why do you think the Pi Zero would do it better? 

 

edit:

That is such expensive DVI to HDMI cable?

 

Edited January 17, 2021 by Simius

[Mazzspeed]

7 minutes ago, Simius said:

 

If you want 80-column hardware solution working with a special handler, you just need an existing, genuine Atari XEP80 device. If you want to convert a color image generated by the GTIA to the digital image, you don’t need any Pi Zero nor an adapter for $29, but an ordinary S-Video cable for a few bucks. Your LCD TV will convert the image itself. Why do you think the Pi Zero would do it better? 

Crosstalk and interference caused by the cable itself as well as increased resistance along the length of the cable. Such things aren't marketing speak like OFC speaker cable, these are measurable and observable phenomenon.

 

There's absolutely no negatives to shifting the scaler into the machine itself, especially for such a low price.

Edited January 17, 2021 by Mazzspeed

[Simius]

If anyone would like to try to build a converter which could convert the GTIA generated color image to the digital data, I'm offering my comprehensive help. I have some experience in that matter and even some good looking results:

http://www.atari.org.pl/forum/viewtopic.php?id=12241

I especially recommend post #54 which contains a conclusion. It means - it is possible to do, but impractical and not cheaper at all. 

Edited January 17, 2021 by Simius

[mdivancic]

We have a couple good options for video output to modern devices with the 8-bit line. What is really needed is something like this for the ST line of computers. 

[ivop]

11 hours ago, ClausB said:

Thanks for the link. Yes the 7V timings make more sense. I don't see any mention of the rising edge timing. The schematic shows 15 delay lines, the first of which is color 1 and also the color burst signal. The other colors are delays of that, so they all share the same duty cycle. The first delay line inputs OSC, which is definitely 50%.

 

Consider this: if duty cycle varied then the average chroma voltage would vary, which would appear as luminance signal, causing a change in brightness with each color change. My first years on the Atari were with a monochrome television and I know there was no luminance change with color change. Furthermore, I could see the chroma signal as a ripple on the monochrome image, and see it vary in phase with color.

Apparently, it was me who misunderstood it for years Thanks for the explanation.

[ivop]

6 hours ago, Simius said:

 

If you want 80-column hardware solution working with a special handler, you just need an existing, genuine Atari XEP80 device. If you want to convert a color image generated by the GTIA to the digital image, you don’t need any Pi Zero nor an adapter for $29, but an ordinary S-Video cable for a few bucks. Your LCD TV will convert the image itself. Why do you think the Pi Zero would do it better? 

There's no cable involved between the video out circuit and the LCD TV. The whole video circuitry is skipped, and led directly (or through a buffer) to the Pi Zero. I also think the adapter board could be made a lot cheaper. But, this is not a project I plan working on. Just giving my opinion and possible solutions as to how to capture the color data.

 

The timer stuff should still work, but done differently. Start timer trigger by falling edge of OSC, end timer at rising edge of COL, or something like that.

 

6 hours ago, Simius said:

edit:

That is such expensive DVI to HDMI cable?

My cable was €10,= incl. shipping. You are right, that's not exactly very expensive. I got my Sophia2 through a group buy. Hopefully I can find the motivation to install it this week. Really looking forward to see your crisp image

 

[manterola]

Capturing the phase of the color data seems to be the difficult part. The luminance sounds easy. This following idea might complicate the design but what about something simple and cheap like a pic to do the phase detection? Is it fast enough?

[ivop]

26 minutes ago, manterola said:

Capturing the phase of the color data seems to be the difficult part. The luminance sounds easy. This following idea might complicate the design but what about something simple and cheap like a pic to do the phase detection? Is it fast enough?

I think you need at least 16x the OSC clock, which would be around 57MHz. But possibly the sampling rate is still to low. You'll probably need around 114MHz (Nyquist). Not sure if there are AVRs/PICs that run that fast (120MHz). Basically, you need enough cycles to detect 16 different phases/delays.

 

A lot of MCUs I have encountered in the past have a trigger/timer unit. You set the timer to 0. Trigger it by OSC going down (start of color clock), timer starts counting (120MHz ticks), stop it when COLOR goes high. The resulting timer value is related to the phase. I hope it is obvious by now that with 32 ticks in one OSC cycle it might still be difficult to detect all 16 states (black/colorburst, and 15 colors). You have to tweak the delay pot very carefully. An ever higher sampling rate would be better.

 

Isn't there a simple IC which can feed a clock, and another signal, and determines the phase shift, and present that as a byte on 8 of its pins or something?

 

 

Edit: Perhaps that's why the Beeb version uses an extra CPLD (which is not needed for the Amiga), to determine the color?

Edited January 17, 2021 by ivop

[manterola]

Actually, there was some movement in the RGBtoHDMI project front regarding Atari 8 bits:

https://github.com/c0pperdragon/Amiga-Digital-Video/issues/5

 

[Rybags]

Maybe a better solution could be to just snoop the ANx bus and address/data/CS inputs to GTIA then produce the video by emulation.

[bfollowell]

On 1/13/2021 at 12:12 PM, ivop said:

Existing software that uses the E: handler will automatically use it. That would be my main goal for a cheap 80-column hardware solution. Sophia2 cost me €65,= including shipping with a group order. If I can get the same for $29 US and a Pi Zero, I'd choose that. And no need for expensive DVI to HDMI cables. Which reminds me, I still have to order such a cable

 

Expensive? I guess everyone's idea of expensive is different, but I paid like $9 for a 6 foot DVI to HDMI cable and it works perfectly. That's certainly not what I would call expensive. Why in the world anyone would pay much more than that is beyond me.

 

[bfollowell]

On 1/17/2021 at 7:48 AM, mdivancic said:

We have a couple good options for video output to modern devices with the 8-bit line. What is really needed is something like this for the ST line of computers. 

 

This has already been proven to work with the Atari ST. Some tweaking is still going on but the are making good progress.

 

https://www.exxoshost.co.uk/forum/viewtopic.php?f=13&t=3838

 

[Gunstar]

On 1/13/2021 at 11:45 AM, pixelmischief said:

Did we learn nothing from the VBXE?

I guess not, since there is now Sophia 2 with new GTIA graphic modes as well. But I guess it depends what you think you've learned from VBXE. I still intend to have both myself, VBXE for my XL and Sophia 2 for my 800 (which I alread got and installed, but it had to be returned and I'm waiting on a replacement still).

 

I'll get plenty of use out of the new graphic modes on both, myself, regardless of lack of support for them from others.

[bfollowell]

The one thing I don't like about this project, for the ST or for the 8-bit, is that it only addresses video, not audio. Without having audio injected into the HDMI stream, it's really not a complete system. A lot of people keep saying, "just run the audio in through the extra audio input". I have no idea what type of display they're using, but I haven't seen an extra audio input on any HDMI source on a TV in ten years. I have ten TVs around my house, all ten years old or newer, all Samsung or LG, and not a single one of them has any additional audio inputs for HDMI sources. They all expect the audio to come in through the HDMI stream.

 

11 minutes ago, Gunstar said:

I guess not, since there is now Sophia 2 with new GTIA graphic modes as well.

 

 

I thought the Sophia2 was just an improved Sophia, with DVI and RGB in one unit with an easy way to change the resolution. I had no idea that it added new graphic modes. I'll have to look into that.

 

[BillC]

2 hours ago, bfollowell said:

And no need for expensive DVI to HDMI cables

I've paid $5 or $6 for a DVI to HDMI adapter, which allows the use of any HDMI cable.