IBM/C128 CGA conversation, capture, and VGA display success stories

[Trixter]

• Another Richard Goedeken's design board, this time revised by John Carlsen: http://www.bit-c128.com/ (I have this in my possession but have not yet verified its functionality)

I was finally able to verify this last night. My initial attempts hooking it up as intended, with a GBS-8220 as a scan converter, were very disappointing; the output levels were extremely dim, and attempting to raise them via the GBS-8220 controls resulted in a washed-out image. John Carlsen suggested there might be a double-termination issue on the GBS-8220 somewhere, but I was unable to verify that without trying another scan converter board. Last night, I did just that; I wired up a quick'n'dirty cable to connect the bit-c128 device to my broadcast-rated converter (Sony DSC-1024) and was treated to stable images with nearly perfect colors and levels:



The DSC-1024 prefers separate horizontal and vertical sync, so removal of the "combined sync" jumper on the bit-c128 device was necessary. (By "prefers", I mean "I wasn't sure if it could deal with combined sync and opted not to attempt it".) The images show that the bit-c128 device itself works as advertised, although your scan converter mileage may vary.

The ultimate goal of this is video capture, so next on my list is reproducing Chris Martin's StarTech PEXHDCAP experiments.

 

[Great Hierophant]

I was finally able to verify this last night. My initial attempts hooking it up as intended, with a GBS-8220 as a scan converter, were very disappointing; the output levels were extremely dim, and attempting to raise them via the GBS-8220 controls resulted in a washed-out image. John Carlsen suggested there might be a double-termination issue on the GBS-8220 somewhere, but I was unable to verify that without trying another scan converter board. Last night, I did just that; I wired up a quick'n'dirty cable to connect the bit-c128 device to my broadcast-rated converter (Sony DSC-1024) and was treated to stable images with nearly perfect colors and levels:

View attachment 29302View attachment 29303View attachment 29304

The DSC-1024 prefers separate horizontal and vertical sync, so removal of the "combined sync" jumper on the bit-c128 device was necessary. (By "prefers", I mean "I wasn't sure if it could deal with combined sync and opted not to attempt it".) The images show that the bit-c128 device itself works as advertised, although your scan converter mileage may vary.

The ultimate goal of this is video capture, so next on my list is reproducing Chris Martin's StarTech PEXHDCAP experiments.

Some progress indeed! I guess John Carlsen did something right after all. Of course, the Sony is extremely expensive and unavailable to most people, so hopefully the PEXHDCAP will work out for you. Still, its rather weird that your GBS-8220 would show such poor input when they recommend it on the http://www.bit-c128.com/ page.

 

[Trixter]

Of course, the Sony is extremely expensive and unavailable to most people, so hopefully the PEXHDCAP will work out for you.

Actually, the DSC-1024, DSC-1024G, and DSC-1024HD have been coming down in price. While I bought my DSC-1024 for $1600 in 2005, you can get the same unit used on ebay for about $120. On a lark, I bid on an extremely beat-up DSC-1024HD recently and won it for $20. It works, but it has seen a *LOT* of use and there is a very slight amount of noise in the signal, more than I would consider acceptable for making reference recordings. But, it's perfectly usable for hooking up a VGA monitor to just use a system, or for something silly like hooking up a system to a 55" TV. (The one I purchased in 2005 has no noise problems, as I am the original owner and have never abused it.)

To anyone wanting a DSC series scan converter, they have some pros and cons:

Pros: Interpolates across 4 scanlines in both the input and output stage for clean images without jagged edges. Recognizes nearly every input signal including 15.7KHz "240p" signals from consoles. Extremely high-quality conversion (stable image, no noise). Very adjustable (size, position, chroma, phase, contrast, aperture). Has memory for user presets. Can be rack-mounted. My favorite: Built-in test patterns.

Cons: Does not have any motion-compensation features. Input rates that differ from output rates will have "motion blur" or double images as the input is linearly interpolated to the output. Large, heavy unit. Fan noise is slightly audible. Uses coax inputs, so RCA connectors will need adapter plugs. No HDMI support (the DSC_1024HD supports 1080i through VGA or component).

 

[njroadfan]

The GBS-8220 converters are known for their lousy quality, but the dark video could be another problem. Some game consoles require passive components inline (usually capacitors in the 220uF range to remove a DC offset on the output) with the RGB video lines to brighten up the video, so its not an issue with just that adapter. The Sony's input circuit likely handles DC offset without a problem, but the bargain basement GBS-8220 likely has nothing on the input side.

For those looking for a better scaler with digital video output and a modest price tag, check out this open source project that is "coming soon": http://shmups.system11.org/viewtopic.php?f=6&t=52158&p=1142878#p1142878

 

[Great Hierophant]

At this point, I would suggest simply using the convert board, an SCART-to-Component converter and an Elgato Game Capture HD.

The converter board translates digital RGBI into analog RGB. The Framemeister upscales it to 720p, and should do so quite well. Then you use the Elgato to capture at 720p/60fps.

If you wanted to save $$, you could just get a SCART to Component converter and use the Elgato, but you would have to de-interlace the video in software. 240p is not available progressively over Component, but is over Composite or S-Video with the Elgato.

 

[Trixter]

Still, its rather weird that your GBS-8220 would show such poor input when they recommend it on the http://www.bit-c128.com/ page.

Actually, it turns out my bit-c128 unit is defective; John is working with me to troubleshoot what is going on. Both my original unit and the replacement unit has shown a markedly lower green level in the output. He is looking into the issue and I'll update this thread when I have a solid solution that works for me (I'm currently correcting the color in post-production, but this is not ideal). I'm currently trying to capture output directly using a StarTech USB3HDCAP, which is a rebranded Micomsoft XCAPTURE-1 without the passthrough ports. (I'm not necessarily recommending as a solution just yet, because while it is working okay with CGA to VGA DACs mentioned in this thread, it has some generic VGA captures issues I'm still working out.)

 

[Scali]

On a related note... Can this TTL-to-RGB analog approach also be used for MDA/Hercules?
Potential problem could be that it is 50 Hz, but not quite PAL, with 370 scanlines per field, where PAL has 312.5 scanlines per field.

Then again, for VGA I have a device that can convert any VGA signal to NTSC or PAL composite/s-video (640x400, 640x480, 800x600, 1024x768, and 60 or 70 Hz. It can probably handle more obscure resolutions as well). It may have enough tolerance to even convert an MDA/Hercules signal to NTSC or PAL. And the TTL-to-RGB analog part isn't sensitive to any frequencies, so it may just work

 

[Trixter]

I'm guessing, but it might be possible using CGA DACs already created. MDA has a single TTL pin (7), and also has the same intensity, hsync, and vsync pins (6, 8, and 9 respectively). If you made a cable that took pin 7 from the MDA side and connected it to the R, G, and B pins (3, 4, and 5 respectively), you'd make a "mono cga" cable. Pass through the intensity, hsync, and vsync, and it might work. Vsync and hsync would be different frequencies, obviously, but the homebrew DACs already mentioned in this thread don't have any sort of framebuffer; I thought they just converted the TTL signals to analog.

 

[vwestlife]

I'm guessing, but it might be possible using CGA DACs already created. MDA has a single TTL pin (7), and also has the same intensity, hsync, and vsync pins (6, 8, and 9 respectively). If you made a cable that took pin 7 from the MDA side and connected it to the R, G, and B pins (3, 4, and 5 respectively), you'd make a "mono cga" cable. Pass through the intensity, hsync, and vsync, and it might work. Vsync and hsync would be different frequencies, obviously, but the homebrew DACs already mentioned in this thread don't have any sort of framebuffer; I thought they just converted the TTL signals to analog.

If you have a Tandy 1000RL/SL/TL, then in MDA/Hercules mode it puts the video output on the green pin as well as on pin 7. After some tweaking of the controls, it can almost give a usable image on a CGA monitor:

 

[Scali]

Well, I hope to be able to do video captures of MDA/Hercules. Perhaps I should make a new topic on that.

 

[Trixter]

Actually, it turns out my bit-c128 unit is defective; John is working with me to troubleshoot what is going on. Both my original unit and the replacement unit has shown a markedly lower green level in the output.

I forgot to update this thread with my final opinions on this device: Based on testing two different units of this device, and based on what John told me about how he designed the circuit, I believe his design methodology was flawed and he designed the circuit incorrectly. I believe the lower green levels are a flaw of all bit-c128 units, and will require raising the green level substantially in the scaler you are using. Because I was outputting directly to a compatible device and didn't need a scaler, this wasn't an option for me. As a result, I cannot recommend the bit-c128 device without the use of a scaler to boost the green level.

My next steps: There is a circuit being designed and built by someone over at the Amibay forums, and it promises excellent quality, so I hope to obtain one of those. Another option is to build one of J. Alex's boards, which at least one person has said good things about.

 

[Scali]

I forgot to update this thread with my final opinions on this device: Based on testing two different units of this device, and based on what John told me about how he designed the circuit, I believe his design methodology was flawed and he designed the circuit incorrectly. I believe the lower green levels are a flaw of all bit-c128 units, and will require raising the green level substantially in the scaler you are using. Because I was outputting directly to a compatible device and didn't need a scaler, this wasn't an option for me. As a result, I cannot recommend the bit-c128 device without the use of a scaler to boost the green level.

Thanks for the heads up.
The bit-c128 is pretty expensive (partly because of shipping to Europe). I was thinking of getting one, since the Amibay ones may take a while to emerge.
But if it has issues with the GBS-8200, then it would be of limited use to me, and not worth the money spent.

By the way, have you tried using the bit-c128 with an MDA/Hercules card yet, on the GBS-8200? I am still wondering if that works, and if that would allow us to capture MDA/Hercules video.

 

[Trixter]

By the way, have you tried using the bit-c128 with an MDA/Hercules card yet, on the GBS-8200? I am still wondering if that works, and if that would allow us to capture MDA/Hercules video.

I recall it not working. When I get a CGA solution functional, I'll test again, but since MDA is a 50Hz vertical refresh rate, I'm not hopeful.

 

[Scali]

I recall it not working. When I get a CGA solution functional, I'll test again, but since MDA is a 50Hz vertical refresh rate, I'm not hopeful.

According to the manual, the input is 'auto scan', and supports a range of input frequencies: https://www.manualslib.com/manual/894635/Digital-Systems-Design-Gbs-8200.html?page=3#manual
MDA has a horizontal frequency of about 18.4 MHz, which may or may not work on the CGA/EGA input.
I thought people also used these for 50 Hz C128s and Amigas, or arcade machines...

 

[Trixter]

I never used the GBS; I always had the device connected directly to my micomsoft capture device, which can report on and adapt to a very wide range of frequencies. MDA, connected through the device, did not sense anything it thought was a video signal.

I think you may be confusing the capabilities of a "CGA TTL -> analog VGA 15.7KHz" device, and a gonbes scan converter device. If the former doesn't produce anything usable, the latter has nothing to adapt to. When I connected the former to MDA, I got nothing usable out of it, and that was the extent of my testing. This is likely because:

CGA pinout:

1 Ground
2 Ground
3 Red
4 Green
5 Blue
6 Intensity
7 Reserved
8 Horizontal sync
9 Vertical sync

MDA pintout:

1 Ground
2 Ground
3 Not connected
4 Not connected
5 Not connected
6 High intensity
7 Video
8 Horizontal sync
9 -Vertical sync

bold indicates pins that actually generate video. Because pin 7 on MDA generates video, and that pin is not connected or reserved on CGA, I don't think any single CGA-to-VGA device will convert it. A custom device would need to be made for MDA.

 

[njroadfan]

What I would really want to see is an "all-in-one" CGA/EGA/MDA TTL to RGB convertor. It shouldn't be too hard considering the number of late 1980's monitors that supported all 3 standards (plus had detection circuits to switch between TTL and analog modes).

I haven't read the entire Amibay thread yet, but hopefully that adapter can handle IBM "brown" without a problem.

 

[Scali]

I think you may be confusing the capabilities of a "CGA TTL -> analog VGA 15.7KHz" device, and a gonbes scan converter device.

I'm not confusing anything.
I think the 'analog VGA 15.7 kHz' device is a horrible misnomer. VGA doesn't support 15.7 kHz, so that's nonsense.
What it is is an analog RGB signal with NTSC timings (the same you would get from an NTSC Amiga for example, or from any other NTSC-capable analog RGB device, such as a game console, VCR, laserdisc player or TV tuner. It's what you normally feed into the SCART connector of a TV/monitor).
The unit is 'stupid' and doesn't do anything with timings whatsoever. It just converts RGBI with 2 discrete values (0 and 1) into analog RGB with 3 discrete values (0, 0.5 and 1), overly simplified. So basically it 'composes' the I-signal into the analog RGB signals.
I don't see why it wouldn't work with MDA or any other device that outputs discrete video signals (MDA just doesn't output RGB, so you just have 'MI', a monochrome signal and an intensity signal. Just connect the M output to R, G and B together...).

I was talking about the GBS, which can convert those NTSC timings to something VGA-compatible (something us Amiga guys know as a 'scandoubler' or 'flickerfixer', although the GBS is slightly more advanced and actually buffers and rescans an entire frame, rather than just scanlines).
Since the GBS can also do that for EGA and other ranges, it may just work for MDA/Hercules as well.
So you'd effectively have an MDA->VGA device. You would still need the 'analog RGB' device to convert the mono and intensity signals into two separate analog voltages though, since the GBS expects analog RGB input.

Obviously the pinout is different, but that is trivial to solve. So I figured I didn't have to specifically spell that out, given my intended use (connecting MDA implies using the MDA signals, and rewiring it so the RGBI->analog RGB DAC can use them. Only question is whether or not the GBS can sync. I don't have a DAC yet, so I can't make a working GBS setup at all, neither for CGA nor MDA).

 

[vwestlife]

Obviously the pinout is different, but that is trivial to solve. So I figured I didn't have to specifically spell that out, given my intended use (connecting MDA implies using the MDA signals, and rewiring it so the RGBI->analog RGB DAC can use them. Only question is whether or not the GBS can sync. I don't have a DAC yet, so I can't make a working GBS setup at all, neither for CGA nor MDA).

Note that the Tandy 1000RL/SL/TL models which are CGA/MDA switchable output MDA/Hercules monochrome video on the green pin (pin 4) as well as on pin 7, so in that case no rewiring is necessary.

In fact, a Tandy CM-11 CGA monitor will sync to an MDA signal if you adjust the vertical hold control, although even with the H/V size controls at their extremes, part of the image gets cut off.

 

[SpidersWeb]

Just wanted to pop in because I'm currently playing around with this same stuff at the moment myself.

I'm using a Xilinx XC9572XL to switch signals between EGA/CGA/MDA automatically in to the GBS8220.
That way I can do CGA brownfix and pin remappings and CSync generation (including flipping VSync) all in firmware/HDL.
Tonight I got it going with CGA colour ok - although I need to fine tune the voltage levels and experiment a bit.

I'm using a resistor ladder - 750 ohm and 390 ohm resistors per color but they were chosen because my output voltages are different than direct off the TTL.
Doesn't look too bad.


Tomorrow I need to plug in an MDA signal, set my board to MDA mode, and cross my fingers. Same query regarding "can it sync".

Edit: only just read that amibay link - that looks like the exact same thing I was trying to do - CPLD + 6 bit "DAC" (resistor ladder). Still happy I'm building my own though.

 

[CarlosTex]

Interesting!

So you plan to support EGA/CGA/MDA in one go? How's the output quality so far?

Hope you don't give up on this.